Printer Friendly

Serum bioavailable testosterone: assayed or calculated?

Bioavailable testosterone (BT), [7] circulating testosterone not bound to sex hormone-binding globulin (SHBG), is a biological marker of androgenicity in men. In older men, BT is related to physiologic changes, decreased muscle strength, bone density, and depressed mood (1-5). Serum BT is widely assayed in testosterone replacement prognms in older hypogonadal men (6, 7). BT is usually measured after removal of SHBG from serum by ammonium sulfate precipitation (50% saturation) and calculation of the percentage of non-SHBG-bound testosterone by use of tracer-binding methods (8, 9) or direct measurement of testosterone in supernatant that contains free and albumin-bound testosterone (10). Both free testosterone and BT may be calculated by measuring total testosterone, SHBG, and albumin concentrations in serum and using the equilibrium binding constants of testosterone to SHBG and albumin given in published equations (11).

Calculation of free testosterone by use of the association constants of testosterone for albumin ([K.sub.a] = 3.6 x [10.sup.4] L/mol) and SHBG ([K.sub.s] = 1 x [10.sup.9] L/mol) produced calculated results that corresponded to those obtained by equilibrium dialysis (11), but in some cases (n = 24) led to higher calculated BT than assayed BT. Because several theoretical SHBG association constant values (range of [K.sub.s] values, 0.27 x [10.sup.9] to 1.9 x [10.sup.9] L/mol) have been reported (11-22), we compared assayed BT (ABT) concentrations measured by 2 different laboratories for 2 large populations of hypogonadal men (n = 1421 and 170), and [10.sup.9] apparently healthy men with calculated BT (CBT) concentrations obtained with different testosterone association constants.

Participants and Methods

Total testosterone, BT, and SHBG were assayed in fasting blood samples drawn from untreated and testosterone patch-treated hypogonadal men (group G1; 1421 samples) in laboratory 1 (Emi INSERM 03-37, Centre de Recherches Chirurgicales, CHU Henri Mondor, Creteil, France), from untreated hypogonadal men (group G2;170 samples) in laboratory 2 (Laboratoire d'Hormonologie, Hopital St. Antoine, Paris, France), and from untreated healthy men (group G3; [10.sup.9] samples) in laboratory 1. The healthy participants were recruited in a health center (IRSA, Tours, France). The procedure was in accordance with the Helsinki Declaration, and each participant gave informed consent. Albumin was measured only in group G1. In groups G1 and G3, we performed total testosterone assays by time-resolved fluoroimmunoassay (TR-FIA) after extraction and then chromatography on Celite (23, 24). Interassay CVs were 4.9%, 5.1%, 4.2%, 4.6%, 3.6%, and 2.2%, respectively, for the following concentrations of serum controls: 2.60 (75), 5.20 (150), 8.66 (250), 10.40 (300), 13.86 (400), and 20.80 (600) nmol/L (ng/dL). Compared with gas chromatography-mass spectrometry (GCMS), the correlation coefficient (r) was 0.99, and the equation for the regression curve was as follows: TR-FIA = 1.0644(GCMS) - 3.58 ng/dL. Nonpanmetric paired comparative tests showed no significant difference. In group G2, total testosterone was assayed with a commercial reagent set (Beckman Immunotech; Ref. IM 1087) after solvent extraction. The highest cross-reactivity of the anti-testosterone antibody (used in the reagent set) was 10% with dihydrotestosterone. Interassay CVs were 5%-10%. To measure BT in serum samples from the 3 groups, we added minute doses of freshly purified tritiated testosterone to serum at 37[degrees]C; we then precipitated the SHBG by adding a saturated ammonium sulfate solution previously equilibrated at 37[degrees]C (8,9). Samples were immediately centrifuged for 15 min at 30008 in a Jouan KR 422 centrifuge previously equilibrated at 37[degrees]C. Duplicate 0.2-mL amounts were taken from the supernatant and placed in minivials. Ultima Gold scintillating fluid (Packard) for radioactivity counting (Tricard 2300 TR; Packard) was added to all vials. We deducted the percentage of non-SHBG-bound tritiated testosterone, or BT, from the radioactivity measurements and calculated the concentrations of serum BT by multiplying %BT by the serum total testosterone concentrations. The interassay CVs of %BT calculated from 3 serum pools measured in each assay run were 4.5%, 3.8%, and 10%, respectively, for mean serum %BT values of 48.9%, 30.4%, and 18.9% (n = 50 runs). SHBG was measured in all sera by an RIA method (SHBG-RIACT reagent set) purchased from Cisbio International/ Schering. Interassay CVs were 7.6%, 8.5%, and 5.9% for serum SHBG at mean concentrations of 13, 38, and 72 nmol/L, respectively (n = 45 runs).

We compared results obtained with the SHBG-RIACT reagent set with those obtained with the DELFIA SHBG (Ref. A070-101; Wallac) and the SHBG IRMA ORION (Ref 68563; Orion Diagnostica). The correlation coefficients (r) were 0.983 and 0.985, respectively (n = 40 in duplicate for each comparison). There was no significant difference between the paired SHBG-RIACT and DELFIA SHBG results or between paired SHBG-RIACT and ORION IRMA results. Moreover, although BT measurement with ammonium sulfate precipitation is widely used, we checked the efficiency of the separation of albumin from SHBG after SHBG precipitation with 50% saturated ammonium sulfate. We assayed albumin concentrations in the supernatant after SHBG precipitation and centrifugation, and in the serum sample. We performed the assays with a nephelometric method with a kinetic reaction on the Array Beckman Analyzer. The albumin concentrations in the supernatant (obtained by adding 1 volume of saturated ammonium sulfate to 1 volume of serum) were one half the concentrations of albumin in the pure serum samples, indicating that no albumin had been precipitated in the assay conditions. The assays were carried out on 12 samples.

We also assayed SHBG concentrations (IRMA; Cisbio International/Schering) in the supernatant after adding saturated ammonium sulfate to the serum samples and performing centrifugation at 37[degrees]C for 15 min at 30008. The assays were performed on 12 samples from group G1, whose SHBG concentrations were 8.5-39 nmol/L. We found no detectable SHBG in the supernatants of the 12 samples. In addition, an assay was performed on the serum of a diethylstilbestrol-treated patient with prostate adenocarcinoma whose SHBG concentration was very high (310 nmol/L). After addition of the saturated ammonium sulfate and centrifugation, we measured the SHBG concentration in the undiluted supernatant and in 1:2,1:4, 1:8,1:16, and 1:32 dilutions of the supernatant. SHBG was undetectable in the undiluted and diluted supernatant samples (the detection limit of the method was <0.5 nmol/L), indicating that no SHBG was present and that, consequently, all of the SHBG had been precipitated by ammonium sulfate. We assayed albumin in samples from group G1 with the bromcresol green dye-binding method on a Hitachi 911 automated analyzer, whereas we considered albumin concentrations in groups G2 and G3 to be constant and equal to 43 g/L in each serum sample.

Using the assay results for total testosterone, SHBG, and albumin, we determined the CBT according to the formulas of Vermeulen et al. (11), applying various association constants ([K.sub.s]) between 0.6 x [10.sup.9] L/mol and 2 x [10.sup.9] L/mol and various [K.sub.a] values. We then compared the CBT with the ABT concentrations.

We based our comparison of ABT and CBT on calculation of the correlation coefficients between CBT and ABT, the CBT/ABT ratio for each sample, and the number of samples for which CBT differed from ABT by less than 10%, 20% and 30%. For this purpose, we determined the (CBT-ABT)/ABT ratio (negative, positive, and absolute ratio), termed the relative difference (RD). These comparisons were carried out with Microsoft Excel software. We also performed global variance analysis and post-ANOVA Bonferroni/Dunn tests to compare ABT and CBT results.


We compared CBT concentrations calculated with the theoretical association constants [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol (11) with ABT concentrations. In group G1 (1421 samples), the correlation coefficient between CBT and ABT was 0.9720 and the mean (SD) CBT/ABT ratio was 1.576 (0.292), indicating that the CBT was higher than the ABT. The number of samples displaying absolute RDs <0.10, <0.20, and <0.30 were only 45, 107, and 204, respectively. Thus, 1217 (1421 - 204) samples (85.6%) exhibited an absolute RD between CBT and ABT greater than 0.30. These results are reported in Table 1. We obtained similar results for group G2 (170 samples): r = 0.9575; CBT/ABT = 1.85. The numbers of samples with an absolute RD <0.10, <0.20, and <0.30 were 8, 14, and 27, respectively. In the group of 109 healthy persons (G3), the correlation coefficient was 0.9312, the CBT/ABT ratio was 1.4945, and the number of samples with an absolute RD <0.10, <0.20, and <0.30 were 0, 9, and 24, respectively. For all 3 groups, CBT values obtained with the theoretical association constants were higher than the ABT values.

For [K.sub.a] = 3.6 x [10.sup.4] L/mol, the number of CBT values that were nearly identical to the ABT values increased when the [K.sub.s] value increased (Fig. 1). For [K.sub.s] = 2.9 x [10.sup.9] L/mol, the absolute RDs were <0.10, <0.20, and <0.30, respectively, for 560, 995, and 1243 samples, corresponding to 39.4%, 70%, and 87.4% of the 1421 samples. The absolute RD decreased for higher [K.sub.s] values, and there were paired optimal [K.sub.s] (2.9 x [10.sup.9]) and [K.sub.a] (3.6 x [10.sup.4]) values for which close correspondence of CBT to ABT values was maximal. In group G2, the optimal [K.sub.s] was 3 x [10.sup.9] L/mol for a [K.sub.a] of 3.6 x [10.sup.4] L/mol. With these paired optimal [K.sub.s] and [K.sub.a] values, 40%, 68%, and 85% of the 170 samples from group G2 had an absolute RD <0.10, <0.20, and <0.30, respectively.


For [K.sub.s] = 1 x [10.sup.9] L/mol in group G1, the number of CBT values nearly identical to the ABT values increased when the [K.sub.a] increased, and an optimal value of [K.sub.a] was reached for [K.sub.a] = 1.1 x [10.sup.4] L/mol (Fig. 2). For [K.sub.a] values >1.1 x [10.sup.4] L/mol, the number of CBT values nearly identical to the ABT values decreased.

We determined the optimal [K.sub.a] values in group G1 for several published [K.sub.s] values (0.6 x [10.sup.9] to 1.9 x [10.sup.9] L/mol; Table 1) and the absolute, negative, and positive RDs, the r values, and the CBT/ABT ratios [mean (SD)] for each optimal pair of [K.sub.s] and [K.sub.a] association constants. The optimal correlation coefficient (r) was 0.9790-0.9792, the CBT/ABT ratio was 1.0136-1.0404, and the numbers of samples with an absolute RD <0.10, <0.20, and <0.30 were maximal and nearly the same regardless of the optimal paired [K.sub.s] and [K.sub.a] values (Table 1). Thus, for an absolute RD <0.30, the number of CBT values differing by <30% from ABT values was 1242-1245. In comparison, with the theoretical association constants, the number of samples with CBT values that differed by <30% from the ABT values was only 204 (Table 1). Moreover, for the same samples, the CBT values did not change regardless of the optimal paired association constants [K.sub.s] and [K.sub.a] used for calculation.


We obtained similar results for group G2 with optimal paired [K.sub.s] and [K.sub.a] values ([K.sub.s] = 0.6 x [10.sup.9] to 1.9 x [10.sup.9] L/mol, corresponding to a [K.sub.a] of 0.5 x [10.sup.4] to 2 x [10.sup.4] L/mol). The percentages of samples with an absolute RD <0.10, <0.20, and <0.30 were 35.2%-37.6%,66.4%-69.4%, and 86.4%-87.5%, respectively. These percentages are close to those reported for group G1. For [K.sub.s] = 1 x [10.sup.9] L/mol, the optimal [K.sub.a] was 1.1 x [10.sup.4] L/mol, as in group G1 (r = 0.9597-0.9610; CBT/ABT ratios = 0.96-1.0).

On the basis of the correlation coefficients, optimal [K.sub.s]/[K.sub.a] pairs were those yielding the greatest correlation coefficient (r = 0.9793). For this correlation coefficient, in group G1, the optimal paired [K.sub.s] and [K.sub.a] values were 0.6 x [10.sup.9] and 0.8 x [10.sup.4] L/mol, 0.8 x [10.sup.9] and 1 x [10.sup.4] L/mo1,1 x [10.sup.9] and 1.4 x [10.sup.4] L/mol, 1.2 x [10.sup.9] and 1.6 x [10.sup.4] L/mol, 1.8 x [10.sup.9] and 2.4 x [10.sup.4] L/mol, and 1.9 x [10.sup.9] and 2.6 x [10.sup.4] L/mol. Although these optimal [K.sub.s] and [K.sub.a] pairs were not exactly the same as those obtained from the RD determination (Table 1), this optimization approach led to practically the same results as those obtained by counting samples with absolute RD values <0.10, <0.20, and <0.30. In group G1, the optimal [K.sub.s]/[K.sub.a] pairs that yielded a CBT/ABT ratio as close as possible to 1 were 0.6 x [10.sup.9] and 0.6 x [10.sup.4] L/mol, 0.8 x [10.sup.9] and 0.8 x [10.sup.4] L/mol, 1 x [10.sup.9] and 1.1 x [10.sup.4] L/mol, 1.2 x [10.sup.9] and 1.3 x [10.sup.4] L/mol, 1.4 x [10.sup.9] and 1.6 x [10.sup.4] L/mol, 1.6 x [10.sup.9] and 1.8 x [10.sup.4] L/mol, 1.8 x [10.sup.9] and 2 x [10.sup.4] L/mol, and 1.9 x [10.sup.9] and 2.2 x [10.sup.4] L/mol, which were very similar to those reported in Table 1.

Whatever the mode of optimization, we found that use of optimal paired [K.sub.s] and [K.sub.a] values yielded a greater number of samples with the CBT close to the ABT, in contrast to the results obtained with the theoretical [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol, as illustrated in Fig. 3.

The calculated CBTs and the corresponding ABTs are shown on the same axis in Fig. 3. The CBTs were calculated based on 2 paired [K.sub.s] and [K.sub.a] values: [K.sub.s] = 1 x [10.sup.9] L/mol with [K.sub.a] = 1.10 x [10.sup.4] L/mol (one of the optimal [K.sub.s]/[K.sub.a] pairs; Table 1), and [K.sub.s] = 1 x [10.sup.9] L/mol with [K.sub.a] = 3.6 x [10.sup.4] L/mol [association constants applied by Vermeulen et al. (11)]. The results show that the CBT values obtained from the association constants of Vermeulen et al. (11) were well above the CBT obtained from optimal paired [K.sub.s] and [K.sub.a] values.


Using global variance analysis and a post-ANOVA Bonferroni/Dune test, we found a significant difference between ABT and CBT values obtained with the formulas of Vermeulen et al. (11) but no significant difference between ABT and one set of the optimal paired [K.sub.s] and [K.sub.a] values reported in Table 1.

In group G3, for the same [K.sub.s], the optimal [K.sub.a] values were higher than in groups G1 and G2 (Table 2). The absolute percentage of CBT values obtained from each pair of optimal [K.sub.s] and [K.sub.a] values that differed by <30% from the ABT values was 97.5% (Table 2). In this group, the mean %BT of the 34 young men (20-39 years of age) was 39% for a mean total testosterone of 16.8 nmol/L and a mean ABT of 6.38 nmol/L.


Vermeulen et al. (11) reported that free testosterone concentrations could be calculated with the association constants [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol. The authors noticed, however, that CBT concentrations were 23-fold and ABT concentrations 20-fold higher than free testosterone concentrations, indicating higher CBT than ABT in a small group of individuals (n = 24) (11). Using the same association constants, we also found, in a rather large sample group of 1421 samples (group G1), that the choice of these 2 numeric values ([K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.s] = 3.6 x [10.sup.4] L/mol) yielded CBT results that were clearly much higher than the ABT results Even higher or similar mean CBT/ABT ratios have also been reported by others [CBT/ABT = 2.2 in 700 patients (Dechaud et al., unpublished data) and CBT/ABT =1.50 in control and hypogonadal men (Tremblay et al., unpublished data)].

Although different values of [K.sub.s] have been published (11-22), we found that by recalculation of the CBT to obtain a greater number of samples with CBT results close to the ABT, a [K.sub.a] of 3.6 x [10.sup.4] L/mol led to optimal [K.sub.s] values of 2.9 x [10.sup.9], 3 x [10.sup.9], and 2.3 x [10.sup.9] L/mol for our sample groups G1, G2, and G3, respectively. The [K.sub.s] values of 2.9 x [10.sup.9] and 3 x [10.sup.9] L/mol, however, were higher than the upper value of the previously published [K.sub.s] value of 1.9 x [10.sup.9] L/mol (11-22). On the basis of the hypothesis that [K.sub.a] = 3.6 x [10.sup.4] L/mol was not the exact association constant of testosterone for albumin in serum, our recalculation of the optimal paired [K.sub.s] and [K.sub.a] values for various [K.sub.s] values of 0.6 x [10.sup.9] to 1.9 x [10.sup.9] L/mol showed that the corresponding [K.sub.a] values were 0.60 x [10.sup.4] to 2.29 x [10.sup.4] L/mol (group G1), 0.5 x [10.sup.4] to 1.97 x [10.sup.4] L/mol (group G2), and 0.8 x [10.sup.4] to 3 x [10.sup.4] L/mol (group G3), lower than the [K.sub.a] (3.6 x [10.sup.4] L/mol) applied by Vermeulen et al. (11).

In group G2, the optimal paired [K.sub.s] and [K.sub.a] values were slightly different from those obtained in group G1, as were the correlation coefficients and CBT/ABT ratios. These slight differences may be related to the methods used to assay total testosterone, which were not identical, and to the choice of the same arbitrary concentration of albumin (43 g/L) in all G2 patients for the calculation of BT. However, the percentages of samples with an RD <0.10, <0.20, and <0.30 were practically the same in groups G1 and G2. In group G3, we observed for one theoretical published K (1.9 x [10.sup.9] L/mol) (12) that the corresponding optimal [K.sub.a] (3 x [10.sup.4]/mol) was only a little lower than the theoretical [K.sub.a]. It is probable that the theoretical [K.sub.a] (3.6 x [10.sup.4] L/mol) determined on pure human albumin (25) is higher than the true [K.sub.a] in serum, which could partly explain an ABT lower than the CBT. Vermeulen et al. (11) hypothesized the presence of lipids to explain why CBT was lower than ABT, and free fatty acids in serum have been reported to change albuminbound steroids (26). It is possible that the higher optimal [K.sub.a] determined in samples from healthy men (group G3) compared with groups G1 and G2 could be attributable to lower serum concentrations of free fatty acids. The mean percentage of BT that we found in 34 young men (20-39 years of age) among the [10.sup.9] healthy men of group G3 was 39%. This value was between the extreme mean values for BT reported previously (20%-50%) in different, rather small populations of young healthy men (10, 27-32), which were obtained by similar, but not strictly identical, BT assay methods (method differences concerning incubation temperature, use or not of a tritiated testosterone tracer, and purification of tritiated testosterone, frequently not reported in the BT assay methods). However, this dispersion in the published assayed percentage of BT of healthy young men does not explain the much lower ABT we measured in the 3 populations compared with the CBT obtained with [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol.

Recently, Emadi-Konjin et al. (33). applying the formulas given by Vermeulen et al. (11), with [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol, found systematic differences between CBT and ABT in samples from a group of almost 400 men. These authors reported "implausibly" higher CBT than ABT, and most of the percentage CBT values were in the 30%-70% range, whereas the corresponding measured %BT results were in the range 10%-40%. On the basis of the best correlation coefficients, these authors empirically adjusted the K and [K.sub.a] association constants and found optimal paired K and [K.sub.a] values of 1.4 x [10.sup.9] L/mol and 1.3 x [10.sup.4] L/mol, respectively. These reported results (33) can be compared with ours: for the same [K.sub.s] (1.4 X [10.sup.9] L/mol), we found optimal [K.sub.a] values of 1.60 x [10.sup.4] L/mol in group G1,1.40 x [10.sup.4] L/mol in group G2, and 2.1 x [10.sup.4] L/mol in group G3. We do not think that lower concentrations of ABT compared with CBT (with the association constants [K.sub.s] = 1 x [10.sup.9] L/mol and [K.sub.a] = 3.6 x [10.sup.4] L/mol) that we and others (Dechaud et al. and Tremblay et al., unpublished data) have found can be explained by a methodologic problem, although differences in methodologies exist. To our knowledge, no such comparisons of CBT and ABT [except by Emadi-Konjin et al. (33)] in large numbers of patients and healthy men have been reported. The CBT largely depends on the [K.sub.s] and [K.sub.a] values chosen. Numerous theoretical [K.sub.s] values have been reported in the past, and the exact [K.sub.s] and [K.sub.a] values in serum are not well known. Moreover, as suggested recently (34, 35), the [K.sub.s] value could vary with age.

In conclusion, by calculating optimal pairs of [K.sub.s] and [K.sub.a], we were able to determine CBT values that better agreed with the ABT values than CBT values determined with theoretical association constants. Using optimal [K.sub.s]/[K.sub.a] pairs, we found in our population of untreated and treated hypogonadal nonfasting patients that 30% of the CBT results differed from ABT by at least 20%, whereas in the population of fasting healthy men, CBT obtained with optimal [K.sub.s]/[K.sub.a] pairs led to CBT values close to the ABT values in 97% of samples. Considering the uncertainty of calculating BT, ABT obtained with ammonium sulfate precipitation seems to be a better method than CBT. It would be wise, however, to thoroughly standardize the BT ammonium sulfate precipitation assay method and to determine BT reference values in men.

We thank Dr. Gerard Brilman (IUT Cachan) for his assistance in using the EXCEL software; Professor Jacques Callebert (Hopital Lariboisiere, Paris) for his expertise in statistics; Dr. Noah Hardy, an American scientist, for assistance with correcting the English in the manuscript; and Dr. Monique Pressac (Hopital Trousseau, Paris) for her expertise in measuring albumin.


(1.) Guechot J, Vaubourdolle M, Ballet F, Giboudeau J, Darnis F, Poupon R. Hepatic uptake of sex steroids in men with alcoholic cirrhosis. Gastroenterology 1987;92:203-7.

(2.) Pardridge WM. Transport of protein-bound hormones into tissues in vivo. Endocr Rev 1981;2:103-23.

(3.) Guechot J, Loric S, Vaubourdolle M, Chretien Y, Giboudeau J, Poupon R. Effect of protein binding on testosterone extraction by human cirrhotic liver: evidence for a dissociation-limited uptake. J Clin Endocrinol Metab 1989;69:200-3.

(4.) Van Den Beld AW, De Jong FH, Grobbee DE, Pols HAP, Lamberts SWJ. Measures of bioavailable serum testosterone and estradiol and their relationships with muscle strength, bone density, and body composition in elderly men. J Clin Endocrinol Metab 2000; 85:3276-82.

(5.) Barrett-Connor E, Von Muhlen DG, Kritz-Silverstein D. Bioavailable testosterone and depressed mood in older men: the Rancho Bernardo Study. J Clin Endocrinol Metab 1999;84:573-7.

(6.) Hajjar RR, Kaiser JE, Morley JE. Outcomes of long-term testosterone replacement in older hypogonadal males: a retrospective analysis. J Clin Endocrinol Metab 1997;82:3793-6.

(7.) Sih R, Morley JE, Kaiser FE, Perry HM, Patrick P, Ross C. Testosterone replacement in older hypogonadal men: a 12-month randomized controlled trial. J Clin Endocrinol Metab 1997;82: 1661-7.

(8.) Tremblay RR, Dube JY. Plasma concentrations of free and non-TeBG bound testosterone in women on oral contraception. Contraception 1974;10/6:599-605.

(9.) Loric S, Guechot J, Duron F, Aubert P, Giboudeau J. Determination of testosterone in serum not bound by sex-hormone-binding globulin: diagnostic value in hirsute women. Clin Chem 1988;34: 1826-9.

(10.) Dechaud H, Lejeune H, Garoscio-Cholet M, Mallein R, Pugeat M. Radioimmunoassay of testosterone not bound to sex-steroid-binding protein in plasma. Clin Chem 1989;35:1609-14.

(11.) 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.

(12.) Burke CW, Anderson DC. Sex hormone binding globulin is an estrogen amplifier. Nature 1982;240:38-40.

(13.) Lutz RA, Lutz-Ewan L, Weder HG. Further studies on the temperature dependence of the binding of testosterone to human pregnancy plasma proteins. Steroids 1973;21:423-31.

(14.) Rosner W, Smith RN. Isolation and characterization of the testosterone- estradiol-binding globulin from human plasma: use of a novel affinity column. Biochemistry 1975;14:4813-20.

(15.) Vermeulen A. Transport and distribution of androgens at different ages. In: Martini L, Motta M, eds. Androgens and antiandrogens. New York: Raven Press, 1997:53-65.

(16.) Nisula BC, Dunn JF. Measurement of the testosterone binding panmeters for both testosterone-estradiol binding globulin and albumin in individual serum samples. Steroids 1979;34:771-91.

(17.) Dunn JF, Nisula BC, Rodbard D. Transport of steroid hormones: binding of 21 endogenous steroids to both testosterone-binding globulin and corticosteroid-binding globulin in human plasma. J Clin Endocrinol Metab 1981;53:58-68.

(18.) Pugeat M, Dunn JF, Nisula BC. Transport of steroid hormones: interaction of 70 drugs with testosterone-binding globulin and corticosteroid-binding globulin in human plasma. J Clin Endocrinol Metab 1981;53:69-75.

(19.) Siiteri PK, Murai JT, Hammond GL, Nisker JA, Raymoure WJ, Kuhn RW. The serum transport of steroid hormones. Recent Prog Horm Res 1982;38:457-503.

(20.) Sodergard R, Backstrom T, Shanbhag V, Carstensen H. Calculation of free and bound fractions of testosterone and estradiol-170 to human plasma proteins at body temperature. J Steroid Biochem 1982;16:801-10.

(21.) Strel'chenok OA, Survilo LI, Tsapelik GZ, Sviridov OV. Purification and physicochemical properties of the sex steroid-binding globulin of human blood plasma. Biokhimiia 1983;48:756-62.

(22.) Westphal U. Steroid-protein interactions II. Monogr Endocrinol 1986;27:1-603.

(23.) Fiet J, Gosling JP, Soliman H, Galons H, Boudou P, Aubin P, et al. Hirsutism and acne in women: coordinated radioimmunoassays for eight relevant plasma steroids. Clin Chem 1994;40:2296-305.

(24.) Fiet J, Giton F, Ibrahim F, Valleix A, Galons H, Raynaud JP. Development of a highly sensitive and specific new testosterone time-resolved fluoroimmunoassay in human serum. Steroids 2004;69:461-71.

(25.) Vermeulen A. Testosterone in plasma: a physiopathological study. Verh K Acad Geneeskd Belg 1973;35:95-180.

(26.) Reed MJ, Cheng RW, Beranek PA, Few JD, Franks S, Ghilchik MW, et al. The regulation of biologically available fractions of oestradiol and testosterone in plasma. J Steroid Biochem 1986;24:317-20.

(27.) O'Connor S, Baker HW, Dulmanis A, Hudson B. The measurement of sex steroid binding globulin by differential ammonium sulphate precipitation. J Steroid Biochem 1973;4:331-9.

(28.) Winters SJ, Kelley DE, Goodpaster B. The analog free testosterone assay: are the results in men clinically useful? Clin Chem 1998;44:2178-82.

(29.) Nahoul K, Roger M. Age-related decline of plasma bioavailable testosterone in adult men. J Steroid Biochem 1990;35:293-9.

(30.) Korenman SG, Morley JE, Mooradian AD, Danis SS, Kaiser FE, Silver AJ, et al. Secondary hypogonadism in older men: its relation to impotence. J Clin Endocrinol Metab 1990;71:963-9.

(31.) Morley JE, Patrick P, Perry HM III. Evaluation of assays available to measure free testosterone. Metabolism 2002;51:554-9.

(32.) Tremblay RR, Masse J. Usefulness and limitation of bioavailable testosterone in assessment of androgenicity during the process of aging in men. Aging Male 1999;2:16-21.

(33.) Emadi-Konjin P, Bain J, Bromberg IL. Evaluation of an algorithm for calculation of serum "bioavailable" testosterone (BAT). Clin Biochem 2003;36:591-6.

(34.) Morley JE, Perry HM III. Androgen treatment of male hypogonadism in older males. J Steroid Biochem Mol Biol 2003;85:367-73.

(35.) Haren M, Nordin BEC, Pearce CEM, O'Loughlin P, Chapman I, Morley JE, et al. The calculation of bioavailable testosterone. In: Robaire B, Chemes H, Morales CR, eds. Proceedings of the VII International Congress of Andrology. Englewood, NJ: Medimond Medical Publications, 2001:209-13.


[1] Emi INSERM 03-37, Centre de Recherches Chirurgicales, CHU Henri Mondor, Faculte de Medecine, Creteil, France.

[2] Laboratoire de Biochimie, Faculte de Pharmacie, Paris, France.

[3] Laboratoire d'hormonologie, Hopital St. Antoine, Paris, France.

[4] Laboratoire d'hormonologie, Hopital St. Louis, Paris, France.

[5] Institut Universitaire de Technologie de Cachan, Cachan, France.

[6] Universite Pierre et Marie, Paris, France.

[7] Nonstandard abbreviations: BT, bioavailable testosterone; SHBG, sex hormone-binding globulin; ABT, assayed bioavailable testosterone; CBT, calculated bioavailable testosterone; and RD, relative difference between calculated bioavailable testosterone and assayed bioavailable testosterone.

* Address correspondence to this author at: Centre de Recherches Chirurgicales, Faculte de Medecine, 8 rue du General Sarrail, 94010 Creteil Cedex, France. Fax 33-1-49-81-35-52; e-mail

Received April 7, 2005; accepted December 8, 2005.

Previously published online at DOI: 10.1373/clinchem.2005.052126
Table 1. Comparison between ABT and CBT in 1421 samples from
hypogonadal men (group G1), obtained with 14 optimal [K.sub.s]/
[K.sub.a] pairs and 1 nonoptimal pair. (a)

 Optimal pairs

[K.sub.s], x [10.sup.9] L/mol 0.6 0.7 0.8
[K.sub.a], x [10.sup.4] L/mol 0.6 0.72 0.9
Slope 1.0882 1.0874 1.0913
r 0.9791 0.9790 0.9792
Mean CBT/ABT 1.0220 1.0174 1.0404
SD 0.2019 0.2022 0.2013
-0.1 < RD < 0, n 251 260 233
0 < RD < 0.1, n 315 305 331
[absolute value of RD] < 0.1
 n 566 565 564
 % 39.8 39.8 39.7
-0.2 < RD < 0, n 441 454 402
0 < RD < 0.2, n 557 543 591
[absolute value of RD] <0.2
 n 998 997 993
 % 70.2 70.2 69.9
-0.3 < RD < 0, n 552 569 501
0 < RD < 0.3, n 691 675 743
[absolute value of RD] <0.3
 n 1243 1244 1244
 % 87.5 87.5 87.5

 Optimal pairs

[K.sub.s], x [10.sup.9] L/mol 0.9 1 1.1
[K.sub.a], x [10.sup.4] L/mol 1 1.12 1.3
Slope 1.0890 1.0884 1.0911
r 0.9791 0.9791 0.9792
Mean CBT/ABT 1.0270 1.0233 1.0394
SD 0.2018 0.2021 0.2014
-0.1 < RD < 0, n 242 255 238
0 < RD < 0.1, n 313 310 324
[absolute value of RD] < 0.1
 n 555 565 562
 % 39.1 39.8 39.5
-0.2 < RD < 0, n 433 442 411
0 < RD < 0.2, n 566 556 583
[absolute value of RD] <0.2
 n 999 998 994
 % 70.3 70.2 70.0
-0.3 < RD < 0, n 539 551 509
0 < RD < 0.3, n 704 693 734
[absolute value of RD] <0.3
 n 1243 1244 1243
 % 87.5 87.5 87.5

 Optimal pairs

[K.sub.s], x [10.sup.9] L/mol 1.2 1.3 1.4
[K.sub.a], x [10.sup.4] L/mol 1.42 1.5 1.6
Slope 1.0904 1.0880 1.0867
r 0.9791 0.9790 0.9790
Mean CBT/ABT 1.0353 1.0210 1.0136
SD 0.2016 0.2022 0.2026
-0.1 < RD < 0, n 247 256 262
0 < RD < 0.1, n 321 307 298
[absolute value of RD] < 0.1
 n 568 563 560
 % 40.0 39.6 39.4
-0.2 < RD < 0, n 421 445 466
0 < RD < 0.2, n 572 551 529
[absolute value of RD] <0.2
 n 993 996 995
 % 69.9 70.1 70.0
-0.3 < RD < 0, n 522 557 582
0 < RD < 0.3, n 722 687 662
[absolute value of RD] <0.3
 n 1244 1244 1244
 % 87.5 87.5 87.5

 Optimal pairs

[K.sub.s], x [10.sup.9] L/mol 1.5 1.6 1.7
[K.sub.a], x [10.sup.4] L/mol 1.82 1.9 2
Slope 1.0904 1.0885 1.0874
r 0.9791 0.9791 0.9790
Mean CBT/ABT 1.0356 1.0240 1.0178
SD 0.2017 0.2022 0.2025
-0.1 < RD < 0, n 247 252 262
0 < RD < 0.1, n 320 309 307
[absolute value of RD] < 0.1
 n 567 561 569
 % 39.9 39.5 40.0
-0.2 < RD < 0, n 420 440 452
0 < RD < 0.2, n 574 557 544
[absolute value of RD] <0.2
 n 994 997 996
 % 70.0 70.2 70.1
-0.3 < RD < 0, n 519 551 566
0 < RD < 0.3, n 723 693 677
[absolute value of RD] <0.3
 n 1242 1244 1243
 % 87.4 87.5 87.5

 Optimal pairs pair

[K.sub.s], x [10.sup.9] L/mol 1.8 1.9 1
[K.sub.a], x [10.sup.4] L/mol 2.24 2.29 3.6
Slope 1.0911 1.0885 1.1905
r 0.9792 0.9791 0.9720
Mean CBT/ABT 1.0397 1.0242 1.5759
SD 0.2016 0.2022 0.2925
-0.1 < RD < 0, n 241 253 17
0 < RD < 0.1, n 320 306 28
[absolute value of RD] < 0.1
 n 561 559 45
 % 39.5 39.3 3.1
-0.2 < RD < 0, n 411 441 18
0 < RD < 0.2, n 582 557 89
[absolute value of RD] <0.2
 n 993 998 107
 % 69.9 70.2 7.5
-0.3 < RD < 0, n 509 551 18
0 < RD < 0.3, n 735 694 186
[absolute value of RD] <0.3
 n 1244 1245 204
 % 87.5 87.6 14.3

(a) Shown are the slopes of the regression curves between ABT and CBT,
the correlation coefficients (r), the mean (SD) CBT/ABT ratios, and
the numbers and percentages of individuals with absolute RDs <0.1,
<0.2, and <0.3.

Table 2. Comparison between ABT and CBT in 109 samples from healthy
individuals (group G3), obtained with 15 optimal [K.sub.s]/[K.sub.a]
pairs and 1 nonoptimal pair. (a)

 Optimal pairs

[K.sub.a], x [10.sup.4] L/mol 0.8 1.2 1.5
[K.sub.s], x [10.sup.9] L/mol 0.6 0.8 1
Slope 0.9687 0.9869 0.9797
r 0.9524 0.9521 0.9522
Mean CBT/ABT 0.9500 0.9815 0.9677
 SD 0.1341 0.1367 0.1351
[absolute value of RD] <0.3
 n 106 105 107
 % 97.2 96.3 98.1

 Optimal pairs

[K.sub.a], x [10.sup.4] L/mol 1.7 1.8 2
[K.sub.s], x [10.sup.9] L/mol 1.1 1.2 1.3
Slope 0.9853 0.9748 0.9800
r 0.9521 0.9523 0.9522
Mean CBT/ABT 0.9784 0.9583 0.9682
 SD 0.1363 0.1341 0.1352
[absolute value of RD] <0.3
 n 107 107 107
 % 98.1 98.1 98.1

 Optimal pairs

[K.sub.a], x [10.sup.4] L/mol 2.1 2.3 2.5
[K.sub.s], x [10.sup.9] L/mol 1.4 1.5 1.6
Slope 0.9713 0.9761 0.9801
r 0.9524 0.9523 0.9522
Mean CBT/ABT 0.9516 0.9606 0.9685
 SD 0.1334 0.1344 0.1352
[absolute value of RD] <0.3
 n 107 107 107
 % 98.1 98.1 98.1

 Optimal pairs

[K.sub.a], x [10.sup.4] L/mol 2.6 2.8 3
[K.sub.s], x [10.sup.9] L/mol 1.7 18 1.9
Slope 0.9730 0.9769 0.9803
r 0.9523 0.9523 0.9522
Mean CBT/ABT 0.9548 0.9622 0.9687
 SD 0.1337 0.1345 0.1353
[absolute value of RD] <0.3
 n 107 107 107
 % 98.1 98.1 98.1

 Optimal pairs

[K.sub.a], x [10.sup.4] L/mol 3.1 3.5 3.8
[K.sub.s], x [10.sup.9] L/mol 2 2.2 2.4
Slope 0.9742 0.9803 0.9779
r 0.9523 0.9522 0.9523
Mean CBT/ABT 0.9571 0.9689 0.9641
 SD 0.1340 0.1353 0.1348
[absolute value of RD] <0.3
 n 107 107 107
 % 98.1 98.1 98.1


[K.sub.a], x [10.sup.4] L/mol 3.6
[K.sub.s], x [10.sup.9] L/mol 1
Slope 1.2194
r 0.9312
Mean CBT/ABT 1.4945
 SD 0.2308
[absolute value of RD] <0.3
 n 24
 % 22

(a) Shown are the slopes of the regression curves between ABT and CBT,
the correlation coefficients (r), the mean (SD) CBT/ABT ratios, and
the numbers and percentages of individuals with an absolute RD < 0.3.
COPYRIGHT 2006 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2006 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Endocrinology and Metabolism
Author:Giton, Frank; Fiet, Jean; Guechot, Jerome; Ibrahim, Fidaa; Bronsard, Francoise; Chopin, Dominique; R
Publication:Clinical Chemistry
Article Type:Clinical report
Date:Mar 1, 2006
Previous Article:Effect of sex and assay method on serum concentrations of growth hormone in patients with acromegaly and in healthy controls.
Next Article:Newborn screening for hepatorenal tyrosinemia: tandem mass spectrometric quantification of succinylacetone.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters