CLIN. CHEM. 26/1. 101-106(1980)

CLINICAL CHEMISTRY, Vol. 26, No. 1, 1980 101

Evaluation and Clinical Application of a Dextran-Charcoal Method forMeasuring Bound Testosterone in Human PlasmaRobert K. Tchoiakian, Gerald M. Jones, Barbara M. Sanborn, Luis J. Rodriguez-Rigau, and Gerald G.Reimondo

A method for rapidly determining bound and free testos-terone in human plasma is evaluated and described, inwhich dextran-coated charcoal is used to separate boundand free testosterone. Intra-and inter-assay CV’s were 1.9and 3.8%, respectively. At least 50% of the “bound”testosteronewas associatedwithspeciesthat behavedlike sex steroid hormone-binding globulin on SephadexG-25 chromatography and polyacrylamide gel electro-phoresis.Bound and free testosteroneconcentrationsweredeterminedinnormal men, normal women (duringthemenstrual cycle and during normal pregnancy), and inwomen with various degrees of hirsutism. We concludethatdeterminingthecirculating proportions of bound andfreetestosteroneishelpfulinthediagnosticevaluationofhirsutism in women, and in the investigation of normal andpathologicalconditions where androgen concentrationsmay be altered.

AddItIonal Keyphrases: hirsutism in women . conditionsaffecting androgen concentrations normal values for menand women chromatography on Sephadex 0-25 . elec-trophoresis on polyacrylamide gel . changes during preg-nancy, lactation, menstrual period nature of binding proteirs)

steroids clinical correlation with bound and free

The biological activity of testosterone is a function of theconcentration of “free” testosterone in the blood. This in turnis determined by the interaction of specific binding proteinswith the total circulating testosterone. Testosterone-bindingglobulin, also called “sex steroid hormone-binding globulin”(SHBG), is considered to be the major specific binding proteinin the circulation; other steroid-binding proteins, such ascorticosteroid-binding globulin and albumin, are not specificfor testosterone, thus they contribute less to total testosteronebinding than does SHBG.

Techniques used before measurement to separate free frombound hormone include: treatment of the sample with dex-tran-coated charcoal (1, 2), equilibrium dialysis (3, 4), gel-filtration (1, 5), polyacrylamide gel electrophoresis (6), pre-cipitation with ammonium sulfate (7, 8), passage through acolumn of diethylaminoethyl-cellulose (9), and ion-exchangecolumn chromatography (10). Many of these methods are timeconsuming and thus of limited value for routine clinical use.Furthermore, some techniques for separating free from boundtestosterone may disturb the physiological equilibrium be-tween them, and the binding proteins involved have not beenidentified.

Department of Reproductive Medicine and Biology, The Universityof Texas Medical School at Houston, P.O. Box 20708, Houston, TX77025.

Received July 16, 1979; accepted Oct. 1, 1979.

A relatively simple and rapid method for determining freetestosterone concentrations in clinical samples is needed. Sucha method should identify the proteins that collectively bindtestosterone, including SHBG.

In our laboratory, as in many others, dextran-coated char-coal is used in radioimmunoassays to separate free and anti-body-bound steroids. We decided to utilize similar materialsto separate bound from free testosterone in plasma samples.Furthermore, we have attempted to assess what proportionof the bound component is attributable to SHBG. We reportsuch a method here and have applied it to the determinationof bound and free testosterone in human plasma samples fromvarious types of subjects.

Materials and Methods

Materials

All solvents used in these procedures were of “nanograde”quality (Mallinckrodt, Inc., St. Louis, MO 63147). Non-ra-dioactive testosterone was obtained from Steraloids, Inc.,Wilton, NH 03086, and recrystallized before use. [1,2,6,7-3H)Testosterone (100 kCi/mol; cat. no. 1978-79 TRK-402;Amersham/Searle Co., Arlington Heights, IL 60005) was pu-rified by thin-layer chromatography before use. Diluent wasprepared by dissolving 1 g of bovine gamma-globulin (FractionII; Sigma Chemical Co., St. Louis, MO 63178) in a liter of sa-line (154 mmol/L). Dextran-coated charcoal was prepared bymixing 0.5 g of dextran (T-70; Pharmacia Fine Chemicals, Inc.,Piscataway, NJ 08854) with 0.5 g of charcoal (“Darco G-60”;Matheson Coleman and Bell, Cincinnati, OH 45212), in a literof saline.

Assay Procedure for Determining Free and BoundTestosterone in Plasma

Plasma samples, 25 ftL, from men and women were deliv-ered into 10 X 75 mm assay tubes, and to each was added anequal volume of diluent containing 18 000 dpm of the [3H1-testosterone. The samples were incubated at room tempera-ture for 10 coin, transferred to an ice bath, and 1 mL of dex-tran-coated charcoal suspension was added. The contents ofthe tubes were then vortex-mixed gently and left in ice for 15mm. The samples were centrifuged (2500 rpm, 10 mm), thesupernate containing the bound fraction was decanted into3 mL of p-dioxane in scintillation vials, 10 mL of scintillationfluid was added, and the radioactivity in the vials was countedin a Tricarb Liquid Scintillation Spectrometer, Model 2450(Packard Instrument Co., Downers Grove, IL 60515).

Bound and free testosterone were determined in 20 repli-cates of the same plasma sample in the same assay, to deter-mine intra-assay variation. Furthermore, samples from apooled specimen of plasma from men were incorporated into15 consecutive assays, to determine inter-assay variation.

102 CLINICAL CHEMISTRY, Vol. 26, No. 1, 1980

The percent bound and free, and the concentration of freetestosterone, were calculated as follows:

% bound = (counts in supernate - nonspecific counts)!

(total counts added - machine background counts) X 100

% free = 100% - % bound

Free testosterone, ng/L = (% free X total testosterone

concn, in ng/L)/100

Evaluation of the Method

Sample volume: To establish what sample volume wasappropriate to use for the determination of free and boundtestosterone, we selected four plasma samples: a pooledspecimen, a sample from one man, and samples from twowomen. We measured bound and free testosterone in thesesamples by the above method, using 5 to 50 tL of eachsample.

Sample dilution: To determine the effect of various dilu-ents on the separation of free from bound testosterone, weexamined four different freshly prepared solutions: (a) dilu-ent, consisting of 1 g of bovine gamma-globulin (Fraction II,Sigma) per liter of saline; (b) diluent plus tris(hydroxy-methyl)methylamine (Tris) buffer, 1.21 g/L, pH 7.4 at 4 #{176}C;(c) isotonic saline plus Tris buffer, 10 mmol/L final concen-tration (pH 7.4 at 4 #{176}C);and (d) saline only.

Bound and free testosterone were determined according tothe above-described procedure, except for successive use ofeach of the above solutions, each of which was tested in du-plicate. Two pooled samples of plasma from normal men andtwo samples from normal women were used. Each solution wastested with various volumes of plasma (10 to 100 giL).

Incubation Time and TemperatureTo determine the optimum incubation time and tempera-

ture required for plasma testosterone to equilibrate withadded [3H]testosterone before separating the free from thebound component, we incubated replicate plasma sampleswith 18 000 dpm of [3H]testosterone in 25 zL of diluent for

various intervals at room temperature or at 4 #{176}C,then deter-

mined bound and free testosterone.

Variables in the Incubation with CharcoalDuration: To determine the optimum duration of incuba-

tion for the bound/free separation, we incubated several 25-Lplasma samples with 25 L of diluent containing 18000 dpmof [3H]testosterone for 10 mm at room temperature, added 1mL of dextran-coated charcoal suspension to each sample, andincubated for 4, 8, 12, 16, or 20 mm at 4 #{176}C.

Charcoal concentration: To determine the effect of varyingthe concentrations of dextran-coated charcoal used, we as-sayed samples containing 25 to 500 ig of charcoal for boundand free testosterone.

Sample StabilityIt is not always possible to determine concentrations of free

and bound testosterone immediately after blood withdrawal,so it was essential to investigate the effect of storage upon theirrelative concentrations in plasma samples. We assayed plasmaspecimens from five normal men and five normal women, bothwhen fresh and and on six occasions during storage at -20 #{176}Cfor 244 days. For each determination the specimens werethawed, samples withdrawn, and the remaining specimenre-frozen.

Nature of the Bound Component

Plasma samples (50 L) from males were incubated with50 iL of distilled water containing the 13H]testosterone(36 000 dpm, equivalent to 46.3 pg), with and without 500-fold

molar excess of unlabeled steroid, for 10 mm at room tem-perature. The samples were transferred to an ice bath, 2 mLof dextran-coated charcoal was added, and each sample wasgently vortex-mixed, placed on ice for 15 mm, and then cen-trifuged (10 mm, 1500 rpm, 4 #{176}C).Individual 0.2-mL aliquotsof the supernates were applied to Sephadex G-25 columns,

polyacrylamide gels, or their radioactivity was counted tomeasure bound steroids. The Sephadex G-25 columns (0.5 x7 cm) were eluted with a pH 7.4 eluent containing, per liter,50 mmol of Tris.HCI, 1 mmol of disodium EDTA, and 100 mLof glycerol, and 1-mm fractions of the eluates were counted.Electrophoresis was performed in 0.5 X 10 cm tubes on the(7.5% crosslinked) gels at 2-3 mA per tube (11, 12). This sys-tem separated SHBG from CBG, as judged by migration ofbound [3H]testosterone and [3H]cortisol as well as the relativeeffectiveness of the respective unlabeled steroids to decreasebinding.

Clinical Application of the MethodWe used the method described to determine free and bound

plasma testosterone in samples obtained from the following

types of subjects: 11 normal men, six normal women, 18women with variable degrees of hirsutism [mild, moderate,or severe by a modification of the criteria of Thomas andFerriman (13)], a normal woman at various times during themenstrual cycle, and a normal woman at various times duringpregnancy and lactation. We also measured total testosteronein all the samples and calculated free testosterone concen-trations as described above.

Total Plasma Testosterone Determined in Samplesfrom Men and Women

Sample preparation: One-half milliliter of plasma fromwomen or 0.1 mL of plasma from men was pipetted into 35-mLglass-stoppered conical centrifuge tubes, followed by 0.5 mLof distilled water and 1800 dpm of [1,2,6,7-3H]testosterone(equivalent to 2.5 pg), for use in recovery estimations.

Blanks and controls: Water and “stripped” plasma (fromwomen) blanks were incorporated into each assay. “Stripped”

plasma from women was prepared by adding activated char-coal (25 gIL) to the plasma and vortex-mixing for 30 mm, thenfiltering and centrifuging at 4 #{176}C.The supernatant fluid wasrecentrifuged three times at 4 #{176}C,then filtered through a45-zm filter (Millipore Corp., Bedford, MA 01730). No tes-tosterone was detectable in such stripped plasma. We incor-porated two groups of controls into each assay: six samplesfrom a pool of plasma from women, to determine intra-assayvariability; and nine samples of stripped plasma samples, eachthree receiving a different amount of testosterone, rangingfrom low to high concentrations, to determine assay preci-sion.

Ext raction: Blanks, controls, and all the plasma sampleswere adjusted to pH 9-10 with 100 mmol/L NaOH and ex-tracted twice with 5-mL portions of ether/chloroform (80/20by vol). The respective extracts were combined and washedtwice with 2-mL portions of distilled water, evaporated undernitrogen, then reconstituted in 0.5 mL of methanol. Wheretriplicate assay samples were obtained, three 0.1-mL aliquotswere transferred into individual 10 X 75 mm glass assay tubes,evaporated in a vacuum oven set at 40 #{176}C,and assayed, while

an equal aliquot of the sample was pipetted into a scintillationvial for recovery estimation. Where single assay determina-tions were done on samples, 0.2-mL aliquots were assayed and0.2 mL was used for recovery estimation.

Radioimmunoassay of testosterone: We have previously

reported the specificity of the testosterone antiserum and theprocedure for radioimmunoassay of testosterone (14, 15). Inthe assay of plasma extracts from both men and women thebound/free separation was done with dextran-coated charcoal

60

0

z

0

40

pppp

HFP

MP4*

20

8C

200.

9

d

ib - io - 3’O - 40 -

80

CLINICAL CHEMISTRY, Vol. 26, No. 1, 1980 103

SAMPLE VOLUME (p1)

Figure 1.Boundtestosterone(%) determined on various volumes(dilutions) of four human plasma samplesFP. normal woman; PP. plasma pool; HFP, hirsute woman; MP, normal man

and the supernate containing the bound fraction was decantedinto 3 mL of p -dioxane in scintillation vials, 10 mL of scin-tillation fluid was added, and the radioactivity in the vials wascounted in the liquid scintillation spectrometer. Intra- andinter-assay coefficients of variation were 6 and 9%, respec-tively.

Results

Assay Procedure for Determining Free and BoundTestosterone in Plasma

Although the assay measures bound testosterone, it is moreappropriate to express clinical data on the basis of the calcu-lated free fraction, because this is the biologically activefraction. When we assayed 20 replicates of the same plasmasample concurrently we obtained a mean of 43.3 ± 0.18 (SEM)percent free testosterone, with an intra-assay CV of 1.9%.When plasma controls from a pool of plasma from men wereincorporated into 15 consecutive assays, we obtained a meanof 44.2 ± 0.4 (SEM) percent free testosterone, with an inter-assay CV of 3.8%.

Evaluationofthe Method

Sample volume: Figure 1 shows the percent bound testos-terone determined in various dilutions of four samples ofhuman plasma. The mean value from such dilutions indicatesthat 25 jzL is the smallest volume of plasma sample thatshould be used in the assay, but use of larger sample volumesdoes not significantly change the results for percent bound.

Sample dilution: We saw no significant difference for thepercent bound testosterone on using the four different dilu-ents shown in Figure 2. Therefore, for convenience, we decidedto use the radioimmunoassay diluent preparation similar tothat already in use in our radioimmunoassays.

Incubation Time and Temperature

There was no statistically significant change in the pro-portions of bound or free testosterone from the first minuteto 2 h of incubation at room temperature. One-way analysisof variance and Duncan’s multiple-range test of the data whenvarious incubation intervals at room temperature and at 4 #{176}Cwere compared indicated that the amount of bound steroidmeasured after a 10-mm incubation with [8H]testosterone atroom temperature did not significantly differ from thatmeasured after incubation for 20 to 50 coin at 4 #{176}C.Conve-nience prompted our use of a 10-coin incubation at roomtemperature.

/44

2b - 4O 60 80 i#{243}oSAMPLE VOLUME Ipli

Fig. 2. Effect of use of various diluents on results for boundtestosterone in plasma from (top data) a woman and (bottomdata) a man#{149},RIA diluent; U, Tris buffer, pH 7.4; 0. salIne; 0, RIA diluent + Tris buffer,pH 7.4

Variablesinthe IncubationwithCharcoal

Duration: Binding was greatest between 1 and 3 coin anddecreased gradually up to 60 mm (Figure 3). When these datawere plotted as log percent bound vs. time, the major disso-ciating component exhibited a t112 of about 100 coin. Thevariability in the estimated percent bound during the second10-coin interval was consistently less than during the first 10coin. In addition, the values for nonspecific samples incorpo-rated into each assay were higher during the first 10-coin in-terval. For these reasons, we chose an incubation of 15 coinwith dextran-coated charcoal.

Charcoal concentration: Varying the concentration ofdextran-coated charcoal did not significantly affect the boundor free testosterone concentration, indicating that the charcoaldid not adsorb additional amounts of binding proteins.

Sample Stability

One-way analysis of variance of the data indicated no sig-nificant difference in the proportions of free and bound tes-tosterone in any of the individual samples during the 244-daystorage, nor was any significant difference found when wecompared the percent difference between consecutive deter-minations. Plasma bound testosterone evidently remainsstable, even with repeated freezing (-20 #{176}C)and thawing (+20#{176}C),for at least eight months.

BindingProteinsinPlasma

The bound-hormone supernate of dextran-charcoal treatedplasma was further characterized by Sephadex G-25 exclusionchromatography and polyacrylamide gel electrophoresis.Roughly 50% of the radioactivity in the dextran-coatedcharcoal supernate (bound fraction) of the normal plasmafrom men was bound to macromolecules excluded fromSephadex G25 (Mr >20000) and which migrate like SHBGon electrophoresis. Testosterone in 500-fold molar excesscompeted successfully for the bound [3H]testosterone in bothsystems; cortisol had only a minor effect. The data indicatethat a substantial proportion of the testosterone in the“bound” fraction is associated with SHBG. Further disso-ciation may have occurred during the subsequent fraction-ation procedures, so 50% is a conservative estimate.

Clinical Application of the MethodTable 1 summarizes our total plasma testosterone con-

0

zD0

90

80

70

60

50

40

30

20

MeanRangeSD

Total

382 31.5

n9/L

120

80 39.2

MeanRangeSD

24

582419-692

90

895

3.0

44.0

36

388

Table 1. Total Testosterone and Percentage andConcentration of Free Testosterone in Plasma of

Men and WomenTotal

testosterone,ng/L

11 normal men

Free testosterone

Table 3. Plasma Steroid Concentrations on Days10-19 of a Normal Menstrual Cycle

MeanRangeSD

57302470-9040

1880

Free

testosterone

% ng/LDay ofcycle

10111213

Total

testosterone,ng/L

193218

249285

Six normal womenMean 155Range 133-173SD 15

47.2

39.7-53.14.1

29.627.4-34.7

2.8

ng/L

27201030-4420

940

46

40-535

Estradlol,n9/L

58 14065 146

80 209

92 161

30.0

29.832.3

32.4

34.128.4

30.8

29.330.2

32.2

14 332

15 51916 350

17 23418 31819 300

Pro9es-terone,

ng/L

149129202137177577

8893471

27023553

113147108699697

246

448

22747

7258

100

104 CLINICALCHEMISTRY,Vol. 26, No. 1, 1980

Table 2. Total Testosterone and Percentage andConcentration of Free Testosterone in Plasma of

Hirsute Women

testosterone, Free testosteroneng/L

Mildly hirsute (n = 6)

252-466 26.3-34.4 77-147

39.2 22734.9-42.2 173-262

Moderately hirsute (n = 6)

Severely hirsute (n = 6)MeanRange 673-121 34.5-53.7 280-446SD 191 6.8 63

10 20 30 40 50 60

TIME (Mn)

Fig. 3. Effect of dextran-coated charcoal on the rate of disso-ciation of bound testosterone in plasma samplesEach point representsa meanof six separatedeterminations

centrations, percent free, and concentrations of free testos-terone in normal men and non-hirsute normal women. Table2 shows our results for women with various degrees of hirsu-tism. The percent of free testosterone for normal men (47.2%)was significantly higher than those for normal non-hirsute(29.6%, p <0.01), mildly hirsute (31.5%, p <0.01), and mod-erately hirsute females (39.2%, p <0.05), but did not differsignificantly from that of women with severe hirsutism(44.0%). Values for total plasma testosterone and percent offree testosterone correlated well with the degree of hirsutism.Differences between women grouped according to severity ofhirsutism were significant. However, in some instances wheretotal testosterone concentrations were similar, there weredifferences in the percent of free testosterone and in the de-gree of hirsutism. For example, a patient with mild hirsutismhad 414 ng of total testosterone per liter, 31.3% (130 ng/L) ofwhich was free; another patient with moderate hirsutism had419 ng of total testosterone per liter, with 42.2% (173 ng/L)free. In all the patients we investigated, the concentration offree testosterone correlated better with degree of hirsutismthan did total testosterone concentration.

Menstrual cycle. The concentrations of plama estradiol,progesterone, testosterone, and percent free testosterone were

determined daily from day 10 to day 19 of the menstrual cyclein a normal ovulating woman (Table 3). The length of themenstrual cycle was 30 days and the pre-ovulatory peak of

estradiol was on the 15th day, followed by a progressive in-crease in progesterone, indicating that ovulation occurred onday 16 of the cycle. The percentage of free testosterone re-mained unchanged between days 10 and 19 of the cycle, whilethe total testosterone concentration was highest on day 15,coinciding with the estradiol peak.

Pregnancy. Table 4 summarizes our data on plasma tes-tosterone and the percent and concentration of free testos-terone at various intervals during pregnancy and lactation ina normal woman. Total testosterone increased progressivelyduring pregnancy and the percentage of free testosteronedeclined, while the concentration of free testosterone re-mained relatively unchanged. Immediately after delivery, totaltestosterone decreased, while the percentage of free testos-terone remained low and the concentration of free testosteronedeclined. Total testosterone remained suppressed after threemonths of lactation, but the percentage of free testosteroneincreased. After lactation ended, total testosterone concen-tration and percent free testosterone returned to pre-preg-nancy values.

DiscussionAlthough our method involves use of dextran-coated

charcoal for separating bound from free testosterone, as used

Before pregnancy 242Pregnancy

3 months4 months5 months6 months7 months

372

542

462

678

865

8 months 6969 months 769

Lactation

10 days 1563 months

After lactation106

325

CLINICALCHEMISTRY.Vol. 26. No. 1, 1980 105

Table 4. Total Testosterone and Percentage andConcentration of Free Testosterone at Various

Intervals during a Normal Pregnancy andLactation

Totaltestosterone,

ag/I

SHBG-like protein. Extrapolation of the dissociation rate dataafter exposure to charcoal back to zero time (Figure 3) placesthe estimate at about 63% SHBG-bound steroid in the plasmafrom normal men, indicating that the fraction measured bythis method includes the contribution of SHBG plus addi-tional binding components.

Free testosterone Although our data on the percentage of free testosteroneag/I in plasma of normal men and women are similar to those re-

31.3 76 ported by others (2, 16), our data on total testosterone con-

centrations in the plasma of normal women are lower than18.5 69 those published (2, 16, 17). This reflects the specificity of the

16.5 89 testosterone antibody and of the assay technique used. Cross

16 8 111 reactivity of the testosterone antiserum with 5a-dihydrotes-

177 120 tosterone is 6.6%, with 5a-androstene-3a,17(-diol is 2.2%, andwith other steroids <1% (15). Values for plasma total testos-

15.9 138 terone in women correlated significantly with degree of hir-

14.5 101 sutism, in agreement with other studies (2, 16, 18). In most

13.8 106 cases, measurement of total testosterone sufficed for the di-agnosis of hyperandrogenism (hirsutism, acne) in women.

16.9 26 However, the degree of hirsutism correlated best with the

28.5 30 concentration of free testosterone. The results suggest that

30 2 98 free testosterone concentrations between 75 and 150 ng/L are_______________ associated with mild hirsutism, concentrations between 150

and 300 ng/L with moderate hirsutism, and concentrationsabove 300 ng/L with severe hirsutism.

Total testosterone values in a normal ovulating woman werehighest at the time of the preovulatory estradiol peak, in

agreement with others (19, 20). This is unsurprising, becausetestosterone is the immediate precursor for estradiol synthesisby ovarian follicles. The percentage of free testosterone wasconstant during the menstrual cycle and not altered duringthe changes in estrogen concentrations, in agreement with thefindings of Kim et al. (21). During pregnancy the percent

bound testosterone increased, probably owing to estrogen-induced increases in SHBG (22), while total testosterone

concentrations increased owing to stimulation of ovariansteroidogenesis by choriogonadotropin. Thus, as a result, freetestosterone concentrations remained relatively constantduring pregnancy. During lactation, total testosterone valueswere low, owing to suppression of the hypothalamic-pitu-itary-ovarian axis, while the percentage of free testosteronereturned to pre-pregnancy values as a consequence of de-

by others (2), this method differs in all other specific experi-mental details. We have used this method routinely on about700 different clinical subjects during more than a year. Weconsider the procedure inexpensive, simple, and reproducible(intra-assay and inter-assay coefficients of variation are 1.9and 3.8%, respectively), and it requires only commonplaceradioimmunoassay materials. Furthermore, our results by thismethod correlate well with clinical evaluation of the pa-tients.

Although various sample volumes can be used for deter-

mining free and bound testosterone concentrations, we electedto use the minimum sample volume of 25 /LL for our proce-dure, because it resulted in minimum variation in the pro-portions of free and bound testosterone. In addition, thevarious solutions used to dilute the plasma samples did notaffect the equilibration of PH]testosterone. Consequently,for simplicity we decided to use our standard radioimmuno-assay diluent, of which fresh preparations are always readilyavailable.

Varying the duration of incubation with [3H)testosteronefrom 10 mm to 2 h did not change the proportions of boundand free steroid found. Comparison of the incubation at roomtemperature and at 4 #{176}Cindicates that at least 20 mm of in-cubation at 4 #{176}Cis required to obtain similar data to thatobtained by incubating at room temperature for 10 mm. Thelatter represents a stable state of tracer dilution, as confirmedby assaying various dilutions of plasma obtained from severalindividuals. This finding is critical in that the operator can

perform large or small assays with similar accuracy andwithout worrying about changes in sample stability.

Separation of free from bound testosterone by use of dex-tran-coated charcoal seemed to occur in two stages: (a) re-moval of the free, loosely bound, and rapidly dissociatingtestosterone; and (b) removal of some of the more slowlydissociating testosterone. Therefore, in selecting the intervalfor dextran-charcoal incubation, we chose one when the pro-portions of free and bound testosterone in samples were stable,reproducible, and rapidly determined: 15 mm. Increasing thedextran-coated charcoal concentration added to each sampleby 20-fold, and using a 15-mm incubation did not alter theproportions of free and bound testosterone.

Judging from the migration behavior on polyacrylamide gelelectrophoresis, at least half of what is measured as bound inthe plasma with the charcoal method is associated with an

creased estrogens.

This work was supported in part by Specialized Population Center

Grant 5-P50-HD08338.

References1. Kato, T., and Horton, R., Studies of testosterone binding globulin.J. Clin. Endocrinol. 28, 1160-1168 (1968).2. Rosenfield, R. L., Plasma testosterone binding globulin and indexesof the concentration of unbound plasma androgens in normal andhirsute subjects. J. Clin. Endocrinol. 32, 717-728 (1971).3. Vermeulen, A., and Verdonck, L., Studies on the binding of tes-tosterone to human plasma. Steroids 11, 609-635 (1968).4. Tulchinsky, D., and Chopra, I. J., Estrogen-androgen imbalancein patients with hirsutism and amenorrhea. J. Clin. Endocrinol.Metab. 39, 164-169 (1974).

5. Vermeulen, A., Verdonck, L., Van der Straeten, M., and One, N.,Capacity of the testosterone-binding globulin in human plasma andinfluence of specific binding of testosterone on its metabolic clearancerate. J. Clin. Endocrinol. 29, 1470-1480 (1969).

6. Corvol, P. L., Chramback, A., Rodbard, D., and Bardin, C. W.,Physical properties and binding capacity of testosterone-estradiol-binding globulin in human plasma, determined by polyacrylamidegel electrophoresis. J. Biol. Chem. 246, 3435-3443 (1971).

7. O’Conner, S., Baker, H. W. G., Dulmanis, A., and Hudson, B., Themeasurement of sex steroid binding globulin by differential ammo-nium sulphate precipitation. J. Steroid Biochem. 4, 331-339(1973).

106 CLINICALCHEMISTRY,Vol. 26, No. 1, 1980

8. Heyns, W., and DeMoor, P., The binding of 17 /3-hydroxy-5a-an-drostan-3-one to the steroid-binding a-globulin in human plasma,as studied by means of ammonium sulfate precipitation. Steroids 18,709-730 (1971).

9. Mickelson, K. E., and Petra, P. H., A filter assay for the sex steroidbinding protein (SBP) of human serum. FEBS Lett. 44, 34-40(1974).10. Harman, S. M., and Danner, R. L., Rapid measurement of anindex of testosterone binding to serum binding globulin using ionexchange columns. J. Clin. Endocrine!. Metab. 45, 953-959 (1977).11. Davis, B. J., Disc electrophoresis-lI. Method and applicationto human serum proteins. Ann. N.Y. Aced. Sci. 121, 404-427(1964).12. Elkington, J. S. H., Sanborn, B. M., and Steinberger, E., The effectof testosterone propionate on the concentration of testicular andepididymal androgen binding activity in the hypophysectomized rat.Mo!. Cell. End ocrinol. 2, 151-170 (1975).13. Thomas, P. K., and Ferniman, D. G. P., Variation in facial andpubic hair growth in white women. Am. J. Phys. Anthropol. 15,171-175 (1957).

14. Tcholakian, R. K., Chowdhury, M., and Chowdhury, A., Recoveryof testicular and pituitary functions in adult male rats after cessationof short and long term estradiol treatment. Biol. Reprod. 19,431-438(1978).

15. Rae, P. N., Moore, P. H., Peterson, D. M., and Tcholakian, R. K.,Synthesis of new steroid haptens for radioimmunoassay. Part V.

19-O-Carboxymethyl ether derivative of testosterone. A highly spe-cific antiserum for immunoassay of testosterone from both male andfemale plasma without chromatography. J. Steroid Biochem. 9,539-545 (1978).16. Wiest, W. G., Paulson, J. D., Keller, D. W., and Warren, J. C., Freetestosterone concentration in serum: A method for determination.Am. J. Obstet. Gynecol. 130, 321-328 (1978).

17. Clark, A. F., Marcellus, S., DeLory, B., and Bird, C. E., Plasmatestosterone free index: a better indicator of plasma androgen activity.Fertil. Steril. 26, 1001-1005 (1975).18. Paulson, J. D., Keller, D. W., Wiest, W. G., and Warren, J. G., Freetestosterone concentration in serum: elevation is the hallmark ofhirsutism. Am. J. Obstet. Gynecol. 128,851-857 (1977).

19. Abraham, G. E., Ovarian and adrenal contribution to peripheralandrogens during the menstrual cycle. J. Clin. Endocrinol. Metab.39, 340-346 (1976).20. Dawood, M. Y., and Saxena, B. B., Plasma testosterone and di-hydrotestosterone in ovulatory cycles. Am. J. Obstet. Gynecol. 126,430-435 (1976).21. Kim, M. H., Rosenfield, R. L., and Dupon, C., The effects ofdexamethasone on plasma free androgens during normal menstrualcycle. Am. J. Obstet. Gynecol. 126,982-986 (1976).

22. DeMertogh, R., Thomas, K., and Vanderheyden, I., Quantitativedetermination of sex-hormone-binding globulin capacity in theplasma of normal and diabetic pregnancies. J. Clin. Endocrine!.Metab. 42, 773-777 (1976).