The Para Substituent of S-3-(Phenoxy)-2-hydroxy-2-methyl-N-(4-nitro-3-trifluoromethyl-phenyl)-propionamides Is a Major Structural Determinant of in Vivo Disposition and Activity of Selective Androgen Receptor Modulators

Androgens promote growth, differentiation, and function of male reproductive organs. In addition, androgens play an important role in development and maintenance of extragenital aspects of the male phenotype, including actions on hair growth, skeletal and muscle growth, fat distribution, and deepening of the voice (Mooradian et al., 1987; Hiort, 2002). Physiological roles for androgens in women have been more difficult to define, but studies suggest that androgens have effects on libido, bone mineral density, and lean body mass in women (Davis et al., 1995; Hiort, 2002). The broad spectrum of physiological actions of androgens suggests that a variety of therapeutic applications are possible. However, androgen therapy has been clinically used to a lesser degree than estrogen therapies.

More widely accepted use of androgen replacement therapy has been hampered by the lack of satisfactory androgen formulations. Current steroidal androgen formulations have serious limitations (Bagatell and Bremner, 1996; Nieschlag and Behre, 1998). Oral administration of unmodified testosterone results in low bioavailability due to hepatic inactivation and high variability in testosterone absorption, making acceptable plasma concentrations difficult to achieve. 17-Alkylated testosterones, such as methyltestosterone and fluoxymesterone, are more resistant to hepatic inactivation and provide improved oral bioavailability. However, they are not commonly used due to their hepatotoxicity and low efficacy (Bagatell and Bremner, 1996). Testosterone esters are frequently used in intramuscular formulations. Esterification at the 17-hydroxy position of testosterone makes it hydrophobic, and results in a prodrug that is released gradually from oily drug-containing vehicles at the site of injection. However, such formulations produce fluctuating plasma levels of testosterone and unpleasant side effects (Snyder and Lawrence 1980; Sokol et al., 1982). Skin patches and implants provide better plasma concentration profiles of testosterone. However, skin irritation at the site of patch application and the need for skill for administration and the risk of extrusion in implant delivery limit the usefulness of these formulations. Another common concern regarding steroidal androgen therapy is cross-reactivity of steroidal androgens and their in vivo metabolites with the other steroid hormone receptors, resulting in unfavorable side effects (Bagatell and Bremner, 1996; Bhasin and Bremner, 1997).

The lack of satisfactory steroidal androgen receptor (AR) agonists has piqued interest in development of nonsteroidal AR agonists. Several laboratories, including ours, demonstrated that nonsteroidal AR agonists can be obtained by structural modification of existing nonsteroidal AR antagonists (Dalton et al., 1998; Edwards et al., 1998, 1999; Hamann et al., 1999; Zhi et al., 1999; Dort et al., 2000; He et al., 2002; Yin et al., 2003b). We were the first to report that structural modification of nonsteroidal AR ligands can produce selective androgen receptor modulators (SARMs) with in vivo anabolic and androgenic activities (Yin et al., 2003a). In our early study, we discovered a group of nonsteroidal AR agonists that possess the ability to bind to AR and to induce AR-mediated gene transcription (He et al., 2002). Acetothiolutamide (Table 1) was chosen as a lead compound due to its in vitro AR agonist activity and lack of interaction with other steroid hormone receptors. However, acetothiolutamide showed little in vivo pharmacological activity, resulting from poor pharmacokinetic properties. In vivo metabolism studies showed that the thioether-linkage and B-aromatic ring of acetothiolutamide are susceptible to oxidation, limiting its in vivo efficacy and exposure (Yin et al., 2003c). We designed a series of ether-linked compounds bearing a variety of B-ring substituents in an attempt to avoid metabolic inactivation (Marhefka et al., 2004). Four halogen-substituted analogs that bind the AR with high affinity and stimulate AR-mediated transcription were identified (Table 1). Interestingly, in vivo pharmacological activity was not correlated with in vitro AR binding affinity. We examined the in vivo dose-response relationship and in vivo pharmacokinetics of these four halogen-substituted SARMs in an attempt to delineate whether the observed discrepancy between in vitro and in vivo pharmacological activity was due to differences in intrinsic pharmacological activity or systemic exposure. We then used this data with previous structure-activity relationship studies, molecular modeling studies, and in vivo metabolism studies to identify the most potent, tissue-selective, and orally bioavailable SARM that we have observed to date. The results of these studies are reported herein.

Materials and Methods

Chemicals. The S-3-(4-halophenoxy)-2-hydroxy-2-methyl-N-(4-nitro-3-trifluoromethylphenyl) propionamides, denoted S-1 (4-fluoro), S-9 (4-chloro), S-10 (4-bromo), and S-11 (4-iodo) hereafter, were prepared in our laboratories as described previously (Marhefka et al., 2004). The internal standard for HPLC analysis, a 2,4-difluoro propionamide, and four (4-nitro/cyano phenoxy)-2-hydroxy-2-methyl-N-(4-nitro/cyano-3-trifluromethylphenyl) derivatives, S-19 through S-22, were synthesized using similar procedures. Reagent grade polyethylene glycol 300 (PEG 300) and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (St. Louis, MO). Alzet osmotic pumps (model 2002) were purchased from Durect Corporation (Cupertino, CA). Ethyl alcohol USP was purchased from Pharmco Products, Inc. (Brookfield, CT). HPLC grade acetonitrile and water were purchased from Fisher Scientific Co. (Pittsburgh, PA).

In Vitro Pharmacological Activity. The AR binding affinity of these compounds was determined using an in vitro competitive radioligand binding assay with [³H]mibolerone (MIB) as described previously (Mukherjee et al., 1999). Briefly, increasing concentrations (10^-2-5000 nM) of each ligand were incubated with rat cytosol, a saturating concentration of [³H]MIB (1 nM), and 1000 nM triamcinolone acetonide to prevent interaction of MIB with progesterone receptors at 4°C for 18 h. At the end of incubation, free and bound [³H]MIB were separated using the hydroxyapatite method. IC₅₀ values were determined by computer-fitting the data for each ligand by nonlinear regression analysis (Pharsight, Mountain View, CA). The apparent equilibrium dissociation constant of the inhibitor (i.e., K_i) for each compound was calculated as K_i = K_d× IC₅₀/(K_d + L), where K_d was the dissociation constant of [³H]MIB (0.19 ± 0.01 nM, previously determined by Mukherjee et al., 1999), and L was the concentration of [³H]MIB used in the experiment (1 nM). Binding affinities of the ligands were then compared using the calculated K_i values.

The abilities of the compounds to modulate the transcriptional activity of AR in a cellular context were measured using a cotransfection assay in CV-1 cells, as described previously (Yin et al., 2003b). CV-1 cells (American Type Culture Collection, Manassas, VA) cotransfected with a human AR expression vector, a luciferase reporter vector, and a control β-galactosidase vector were used to evaluate the in vitro functional activity of these compounds. Transcriptional activation was assayed using a single concentration (100 nM) of each ligand and expressed as a percentage of that induced by 1 nM dihydrotestosterone.

Assay for in Vivo Pharmacological Activity in Rats. All animal protocols were reviewed and approved by the Institutional Laboratory Animal Care and Use Committee of The Ohio State University. The in vivo pharmacological activities of these compounds were determined as the increase in weight of target tissues of castrated rats that received the drug of interest for 14 days, as described previously (Yin et al., 2003a). Briefly, immature male Sprague-Dawley rats (Harlan, Indianapolis, IN), weighing 180 to 220 g, were randomly distributed into 23 groups of five animals. The castrated male rats in each group received increasing doses (0.1, 0.3, 0.5, 0.75, or 1 mg/day) of the designated compound via implantation of subdermal osmotic pumps (for S-9, S-10, and S-11) or daily subcutaneous injection (for S-22) for the dose-response analysis. For the in vivo pharmacological activity assay of S-19, S-20, and S-21, the castrated rats received the designated compound at the dose rate of 1 mg/day via daily subcutaneous injections. After 14 days of drug treatment, the ventral prostates, seminal vesicles, and levator ani muscle were removed and weighed. Rats were weighed, anesthetized, and sacrificed. Plasma samples were collected and stored at -20°C. The weights of all organs were normalized to body weight and compared with the historical data of intact control and castrated control animals. The weights of prostate and seminal vesicles were used as markers of androgenic effects and that of levator ani muscle was used as a marker of anabolic effects. The maximal pharmacological effect (E_max) and the dose required to elicit 50% of this effect (ED₅₀) were obtained by nonlinear regression analysis using the simple or sigmoid E_max model in WinNonlin (Pharsight). The relative efficacy of each compound to testosterone propionate (TP) was defined as the ratio of (E_max of the compound) to (E_max of TP). The relative potency was defined as the ratio of (ED₅₀ of TP) to (ED₅₀ of the compound). Differences between groups were determined using ANOVA. P values less than 0.05 were considered as statistically significant. The area under the concentration-time curve (AUC) of compounds in the rats for pharmacodynamic studies was calculated by AUC = dose rate/clearance. The drug clearance after a single intravenous dose (10 mg/kg) was used for this calculation, assuming dose-independent clearance in the range of 0.4 to 10 mg/kg dosing rate for these compounds (Kearbey et al., 2004).

Pharmacokinetic Studies. The pharmacokinetics of four halogen-substituted SARMs and cyano/nitro-substituted SARMs were determined after an intravenous (i.v.) bolus or oral dose at 10 mg/kg body weight. Immature, male Sprague-Dawley rats weighing approximately 250 g were randomly distributed into eight groups (i.v. and oral dose of S-1, S-9, S-10, and S-11) of five animals and four groups (i.v. dose of S-19, S-20, S-21, and S-22) of four animals. A catheter was implanted in the right jugular vein of each animal 1 day before dosing. The compounds were dissolved in a solution consisting of 5% DMSO and 95% PEG 300 and injected as a bolus dose via the jugular vein or administered orally by gavage. The jugular vein catheter was rinsed with saline (three times the volume of the dosing solution) after administration of each intravenous dose. Blood samples (250 μl) were collected via the jugular vein at different predetermined time intervals after the dose. Blood samples were centrifuged at 2000g at 4°C for 10 min, and plasma fractions were stored at -20°C until analysis.

Bioanalytical Methods. An aliquot (100 μl) of plasma from the pharmacokinetic study was placed in a polypropylene tube with 100 μl of an internal standard solution (10 μg/ml solution of the 2,4-difluoropropionamide derivative in acetonitrile for S-1, S-9, S-10, S-11, and S-19; 10 μg/ml solution of S-1 in acetonitrile for S-20, S-21, and S-22) and 800 μl of acetonitrile. The mixture was briefly vortexed and centrifuged at 10,000g at 4°C for 2 min. The supernatant was transferred to another tube and evaporated under nitrogen. The residue was reconstituted with 150 μl of mobile phase, and an aliquot (100 μl) was injected to a C8 column (Symmetry, 3.9 × 150 mm; Waters, Milford, MA). The mobile phase consisted of acetonitrile/H₂O [40:60 (v/v)] for S-1, acetonitrile/H₂O [60:40 (v/v)] for S-9 and S-10, acetonitrile/H₂O [63:37 (v/v)] for S-11, or acetonitrile/H₂O [57:43 (v/v)] for S-19, S-20, S-21, and S-22 at a flow rate of 1 ml/min. Analytes were detected by their UV absorbance at 297 nm (S-1), 282 nm (S-9 and S-10), 233 nm (S-11), 307 nm (S-19), and 272 nm (S-20, S-21, and S-22). The HPLC system consisted of a solvent pump, a degasser, an autosampler, and a UV detector (model 1100; Agilent Technologies, Palo Alto, CA). The accuracy, precision, and recovery of each bioanalytical method were established according to Food and Drug Administration guidelines. The limits of detection were 0.5 mg/l (S-1, S-20, and S-21), 0.3 mg/l (S-9, S-11, S-19, and S-22) or 0.6 mg/l (S-10). The pharmacokinetic parameters were obtained by noncompartmental methods using WinNonlin. The area under the plasma concentration-time curve from time 0 to infinity (AUC_∞) was calculated using the trapezoidal method with extrapolation to time infinity. The plasma clearance (CL) was calculated as CL = dose_i.v./AUC_{∞ i.v.}, where the dose_i.v. and AUC_{∞ i.v.} were the i.v. dose and corresponding AUC_∞ after i.v. administration, respectively. The terminal half-life (t_1/2) was calculated as t_1/2 = 0.693/λ, where λ was the rate constant characterizing the terminal disposition phase. The apparent volume of distribution at equilibrium (V_ss) was calculated as V_ss = CL · MRT, where the MRT was the mean residence time after the i.v. dose. Oral bioavailability (F_p.o.) was defined as F_p.o. = (AUC_∞,p.o. · dose_i.v.)/(AUC_∞,i.v. · dose_p.o.)^-1, where the dose_p.o. and AUC_∞,p.o. were the oral dose and corresponding AUC_∞ after oral administration, respectively.

Results

Halogen-Substituted SARMs

In Vivo Pharmacological Activity of Four-Halogen-Substituted SARMs. The ability of the four-halogen-substituted SARMs to stimulate weight gain of prostate, seminal vesicles, and levator ani muscle in castrated male rats was studied to evaluate their anabolic and androgenic activity. The weights of prostate and seminal vesicles were used as markers of androgenic effects and that of levator ani muscle were used as a marker of anabolic effects. The pharmacological activity of S-1 in these tissues was reported previously (Yin et al., 2003a). Figure 1A shows that S-9 stimulated the growth of prostate, seminal vesicles, and levator ani muscle in castrated rats in a dose-dependent manner. S-9 maximally restored the weights of prostate, seminal vesicles, and levator ani muscle to 33.8 ± 4.0, 28.5 ± 3.9, and 137 ± 9.4% (mean ± S.D.), respectively, of the intact control, compared with 14.5, 12.7, and 74.9%, respectively, as reported previously for S-1. S-9 fully maintained the levator ani muscle weight in castrated animals at the same level as intact control, at a dose rate as low as 0.5 mg/day. At higher dose rates, S-9 promoted growth of the levator ani muscle to a size significantly greater than that observed in the intact control (Fig. 1A). Nonlinear regression analysis of the dose-response relationships for S-9 showed that the ED₅₀ values were 0.25 ± 0.06, 0.48 ± 0.09, and 0.29 ± 0.09 in prostate, seminal vesicles, and levator ani muscle, respectively (Fig. 1A; Table 2). The relative efficacy and potency of S-9 in androgenic and anabolic tissues were compared with those observed in the testosterone propionate-treated group (Table 2). The relative efficacy of S-9 in levator ani muscle was 1.31, which is much higher than its relative efficacies in prostate and seminal vesicles (0.28 and 0.41, respectively), clearly revealing the efficacious and selective activity of S-9 in anabolic tissues of the castrated male rat.

Figure 1B shows the androgenic and anabolic activity of S-10 in castrated rats. S-10 demonstrated the weakest pharmacological activity within this series of compounds, with E_max values of 18.8 ± 3.0, 11.5 ± 0.6, and 64.4 ± 3.7 in the prostate, seminal vesicles, and levator ani muscle, respectively (Table 2). However, S-10 displayed lower ED₅₀ values (0.37 ± 0.25, 0.10 ± 0.04, and 0.11 ± 0.09 in prostate, seminal vesicles, and levator ani muscle, respectively) than the other four-halogen-substituted SARMs, suggesting that higher concentrations of S-10 might have been achieved in the tissues. However, it is important to note that these estimates were also generally more variable and may be low due to the limited response observed with this compound.

Despite its low AR binding affinity and limited in vitro activity (Table 1), S-11 demonstrated promising in vivo pharmacological activity, maintaining the prostate, seminal vesicles, and levator ani muscle in castrated rats at 25.3 ± 2.4, 19.4 ± 2.0, and 95.3 ± 7.2, respectively, of the size of these organs observed in intact controls (Fig. 1C). This result suggested that slower drug metabolism and/or more favorable pharmacokinetic characteristics of S-11 might compensate for its poor AR binding affinity. The relative efficacy of S-11 in androgenic tissues was less than 0.3 compared with TP, whereas its relative efficacy in the levator ani muscle was 0.91, indicating a greater degree of activity in anabolic tissues (Table 2).

Pharmacokinetics of Four-Halogen-Substituted SARMs. Mean plasma concentrations of S-1, S-9, S-10, and S-11 obtained after a single intravenous dose (10 mg/kg) of each SARM are presented in Fig. 2A. Plasma concentrations of these compounds declined rapidly after intravenous administration, plateaued in some cases, and then declined in a monoexponential manner with terminal half-lives ranging from 4 to 14.7 h (Table 3). As the size of halogen atom in B-ring increased, volumes of distribution and clearances decreased. However, clearance values decreased to a greater extent, resulting in prolonged terminal half-lives (i.e., S-11 > S-10 > S-9 > S-1). AUC_∞ was significantly larger for S-10 and S-11, indicating greater systemic drug exposure for these compounds. The AUC_∞ values of S-1, S-9, S-10, and S-11 were 43 ± 5, 160 ± 33, 401 ± 21, and 1227 ± 206 μg · h/ml, respectively. The pharmacokinetics of these compounds after a single oral dose were also examined (Fig. 2B). After oral dosing at 10 mg/kg, the plasma concentrations of S-1, S-9, S-10, and S-11 increased gradually with peak concentrations (2.1 ± 0.2, 6.8 ± 0.8, 11.7 ± 2.9, and 22 ± 4.2 mg/l, respectively) achieved at 4.8 ± 2.2, 9.2 ± 3.3, 9.7 ± 2.4, and 17.4 ± 6.4 h, respectively, after oral administration. Plasma concentrations then declined with similar terminal half-life as that observed after the intravenous dose. Oral bioavailability of S-1, S-9, S-10, and S-11 at the 10 mg/kg dose was 55, 85, 69, and 84%, respectively.

Exposure-Response Relationship for SARMs. In vivo pharmacokinetic data suggested that the observed disparity between in vitro (i.e., AR binding affinity and ability to stimulate AR-mediated transcriptional activation) and in vivo pharmacological activity might be due to differences in systemic exposure. We constructed AUC-response relationships to compare the in vivo activities of the halogen-substituted SARMs at the same level of drug exposure. S-4, which showed the most efficacious in vivo activity in our historical data, also was compared. Figure 3 shows the AUC-response relationship in levator ani muscle. Similar trends in the AUC-response relationship were observed in prostate and seminal vesicle (data not shown). Interestingly, S-4, S-1, and S-9 exerted nearly identical in vivo pharmacological activity when normalized to the same level of drug exposure. S-11, which was more efficacious than S-1 at comparable doses, was significantly less efficacious when compared at the same level of drug exposure. These results suggested that the in vivo pharmacological activity of SARMS with high AR binding affinity (i.e., K_i < 10 nM) was largely governed by in vivo drug exposure, whereas sufficient in vivo exposure to elicit full in vivo agonist activity might be difficult to achieve for SARMs with lesser (i.e., K_i > 10 nM) AR binding affinity.

Cyano/Nitro Group-Substituted SARMs

Binding Affinity and in Vitro Functional Activity of Cyano/Nitro Group-Substituted SARMs. The in vitro and in vivo pharmacological activity and pharmacokinetic studies of halogen-substituted SARMs described above suggested that two simple criteria (i.e., AR binding affinity and in vivo clearance) could be used to select SARMs with promising in vivo pharmacological activity. We tested this hypothesis using a series of four compounds incorporating either a nitro or cyano substituent at the para-position of the A- and B-aromatic rings (Table 4). Previous structure-activity relationship studies, molecular modeling studies, and in vivo metabolism studies suggested that cyano substituents at these positions might be structurally and metabolically favorable (Mukherjee et al., 1999; He et al., 2002; Yin et al., 2003b; Bohl et al., 2004). All of these compounds (S-19, S-20, S-21, and S-22) bound to AR with high affinity and demonstrated agonist activity during in vitro cotransfection studies (Table 4), with no apparent differences to distinguish one compound from another.

Pharmacokinetics of Cyano/Nitro Group-Substituted SARMs. We then determined the pharmacokinetics of S-19, S-20, S-21, and S-22 after administration of a single intravenous dose at 10 mg/kg (Fig. 4). Plasma concentrations of S-19, S-20, S-21, and S-22 declined in a biexponential manner after intravenous administration with terminal half-lives of 4.0, 3.7, 2.6, and 6.0 h, respectively (Fig. 4; Table 4). The V_ss values of S-19, S-20, S-21, and S-22 were 1295 ± 171, 686 ± 42, 834 ± 88, and 635 ± 84 ml/kg, respectively. S-19 and S-21 (CL_S-19 = 4.1 ± 0.1, CL_S-20 = 4.0 ± 0.3 ml/min/kg), compounds bearing a nitro group in the para-position of the B-ring, had higher CL values than S-20 and S-22 (CL_S-20 = 2.4 ± 0.6, CL_S-22 = 1.4 ± 0.3 ml/min/kg), compounds bearing a cyano group in the para-position of the B-ring. Corresponding to its lower CL value, S-22 showed the highest AUC_∞, followed sequentially by S-20, S-21, and S-19. Based on the similar AR binding affinity but lower CL of S-22 compared with S-19, S-20, and S-21, we predicted that S-22 would demonstrate the most promising in vivo activity.

In Vivo Pharmacologic Activity of Cyano/Nitro Group-Substituted SARMs. We compared the in vivo androgenic and anabolic activities of this series of cyano/nitro substituted SARMs at a 1 mg/day dose rate. Figure 5 shows that these compounds significantly stimulated the growth of prostate, seminal vesicles, and levator ani muscle in castrated rats. S-19, S-20, S-21, and S-22 restored the weights of the prostate to 25.2 ± 3.7, 34.2 ± 7.8, 31.2 ± 5.8, and 39.2 ± 8.7%, respectively, of that observed in the intact control. The weight of seminal vesicle was also restored by S-19, S-20, S-21, and S-22 to 19.0 ± 4.8, 47.4 ± 6.8, 21.5 ± 3.1, and 78.8 ± 9.1%, respectively, of that observed in the intact control. S-19, S-20, S-21, and S-22 stimulated the growth of levator ani muscle to a greater extent than androgenic organs, to 105.0 ± 13.4, 130.4 ± 4.9, 118.5 ± 15.8, and 141.9 ± 16.5, respectively. In vivo pharmacological activity correlated well with the AUC_∞ observed in pharmacokinetic studies within this series of cyano/nitro-substituted SARMs (r² = 0.828), confirming that AUC_∞ is useful to predict in vivo activity for SARMs with high (i.e., K_i < 10 nM) AR binding affinity.

Expanded studies of the dose-response relationship for S-22 showed the highest in vivo anabolic activity of any compound that we, or anyone else, have examined. Nonlinear regression analysis of dose-response relationships for S-22 showed that the ED₅₀ values were 0.12 ± 0.05, 0.39 ± 0.15, and 0.03 ± 0.01 in prostate, seminal vesicles, and levator ani muscle, respectively (Fig. 6; Table 2). S-22 maximally restored the weights of prostate, seminal vesicles, and levator ani muscle to 51.1 ± 4.2, 98.0 ± 13.2, and 136.3 ± 3.5%, respectively, of the intact control. S-22 exerted efficacious and selective activity in anabolic tissues at dose rates as low as 0.03 mg/day, indicative of the high potency of this compound in anabolic tissue (relative potency 4.41) and its potential for clinical use.

Discussion

In our early studies, we discovered a group of nonsteroidal AR ligands that possess in vitro AR agonist activity and selected acetothiolutamide as a lead compound. However, acetothiolutamide failed to show in vivo pharmacological activity due to its poor pharmacokinetic properties. Based on in vivo metabolism studies and structure-activity relationships for AR binding, we found that structural modification to an ether linkage instead of a thioether linkage reduced metabolic inactivation. S-4 and S-1 were chosen as second-generation lead compounds. In subsequent studies, we have focused on discovery of SARMs having efficacious and potent in vivo pharmacological activity and favorable pharmacokinetic properties by structural modification of the SARM pharmacophore, focusing mainly on the B-ring since it is amenable to structural modification and critical for agonist activity. Four-halogen-substituted SARMs that displayed potent in vitro and in vivo activity were identified in our previous study (Marhefka et al., 2004). In vitro AR binding affinity of halogen-substituted SARMs increased as the size of the halogen atom in the B-ring decreased and/or electronegativity increased; S-1 > S-9 > S-10 > S-11. Interestingly, in vivo pharmacological activity was not correlated with in vitro AR binding affinity; S-9 > S-11 > S-1 > S-10. Pharmacokinetic studies of the halogen substituted SARMs in the present study provided a valid explanation for this observed disparity, showing that the greater in vivo exposure of drug to target tissue results in greater in vivo pharmacological effect than one would expect based on its AR binding affinity. Moreover, the AUC-response relationships suggest that the observed discrepancy between in vitro and in vivo pharmacologic activity of halogen-substituted SARMs was due to differences in systemic exposure rather than intrinsic pharmacologic activity. Interestingly, the in vivo activity of compounds having binding affinity higher than an empirically determined cut-off (i.e., K_i < 10 nM) was mainly governed by drug exposure (i.e., there were no apparent differences in intrinsic in vivo activity when normalized for exposure). Based on this finding, we hypothesized that two simple factors could be used to identify efficacious and potent SARMs in this chemical class. First, we selected compounds that possess K_i values lower than 10 nM. We then measured the in vivo CL of selected compounds as an additional selection criterion before dose-response analysis. A series of compounds with cyano or nitro substituents at the para-positions of the A- and B-rings were selected for further study and demonstrated high AR binding affinity, suggesting that their in vivo activity would be closely related to their in vivo disposition. Pharmacokinetic studies of the selected compounds showed that the AUC_∞ of S-22 was the largest followed by S-20, S-21, and S-19. In vivo pharmacological studies of these compounds at the dose rate of 1 mg/day displayed the same rank order, as predicted based on the pharmacokinetic study. The dose-response relationship of S-22, displaying the lowest CL among the series of compounds, was examined. In fact, S-22 demonstrated the greatest in vivo androgenic and anabolic activity of any AR nonsteroidal agonist examined to date and exceeded the anabolic activity of TP. Moreover, S-22 was 4 times as potent as TP in levator ani muscle.

The present study demonstrates the importance of in vivo pharmacokinetics and metabolism to predict and improve the in vivo response of SARMs. Structural modification of our SARM pharmacophore resulted in various pharmacokinetic profiles. Despite structural similarities, bicalutamide and S-4 exhibit vastly different pharmacokinetic profiles. Bicalutamide, a clinically used antiandrogen, was eliminated with a half-life of 18 to 21 h in rats (Cockshott et al., 1991). S-4 was eliminated with a much shorter half-life of 3.7 h in rats (Kearbey et al., 2004). The clearance of bicalutamide (CL_Bicalutamide = 0.8 ml/min/kg) was slower than that of S-4 (CL_S-4 = 1.5 ml/min/kg). Bicalutamide (V_{ss Bicalutamide} = 1.2-1.3 l/kg) also has a larger volume of distribution than S-4 (V_{ss S-4} = 0.4 l/kg). S-1 is a structural intermediate with close similarity to both S-4 and bicalutamide. Fluoro substitution of the B-ring, compared with the acetamido group of S-4, produced a larger volume distribution (V_{ss S-1} = 1.5 l/kg) but faster clearance (CL_S-1 = 4.0 ml/min/kg), resulting in a similar half-life (t_{1/2 S-1} = 4 h) to S-4 (t_{1/2 S-4} = 3.7 h). It is interesting to note that S-1 was cleared faster than S-4 despite the expectation that the fluoro substituent in the B-ring would be metabolically more stable than the acetamido group, suggesting that the presence of fluoro substitution on the B-ring might increase the rate of hydroxylation or amide hydrolysis (McKillop et al., 1995; Yin et al., 2003c) and/or increase the overall apparent CL due to an increase in the fraction unbound in plasma. When comparing the pharmacokinetics of S-1 to bicalutamide, which differ in the para-substituent of the A-ring and linkage (i.e., ether versus sulfonyl), S-1 demonstrated a similar volume of distribution but faster clearance (5-fold), resulting in a shorter half-life (5-fold). Subsequent metabolism studies with S-1 (data not shown) suggested that the higher CL of S-1 was due to reduction of the nitro group in the A-ring, suggesting that replacement of the nitro group with a cyano-substituent on the A-ring would reduce clearance and enhance pharmacological activity in vivo. These results demonstrate the significance of A/B-ring substitution on the pharmacokinetic characteristics of SARMs. B-ring substitution at the para-position with halogens significantly affected the volume of distribution and clearance. Compounds incorporating larger halogen substituents (e.g., bromine or iodine) produced slower clearance, resulting in prolonged terminal half-lives in spite of a reduction in the volume of distribution, suggesting that larger halogens and/or less electronegative substituents at this position might decrease the rate of B-ring hydroxylation and/or affect the volume of distribution and CL by altering plasma protein binding. In the series of compounds with cyano- or nitro-substituents at the para-position of the A- or B-ring, the compounds bearing a nitro group in the B-ring (CL_S-19 = 4.1, CL_S-21 = 4.0) were cleared faster than their cyano-substituted counterpart (CL_S-20 = 2.4, CL_S-22 = 1.4). This result suggests that the presence of a nitro group on the B-ring increases drug metabolism via reduction of the nitro group to an amine (Purohit and Basu, 2000). Comparing the two compounds having a cyano group in the B-ring, S-22 (CL_S-22 = 1.4), substituted with a cyano group at the para-position of the A-ring, was more metabolically stable than S-20 (CL_S-20 = 2.4), its counterpart containing a nitro group in A-ring, indicating that the nitro group in A-ring was also more susceptible to in vivo metabolism than the cyano group. In the series of compounds having nitro group in B-ring (S-19 and S-21), incorporation of a cyano group in the A-ring was not helpful to reduce the clearance because S-19 and S-21 showed faster CL than the other compounds, suggesting that the CL induced by nitro-substitution on B-ring was the major factor controlling the rate of metabolism in this class of compounds.

In summary, we examined the in vivo pharmacological activity and pharmacokinetics of halogen-substituted SARMs in the present study. The results demonstrate the importance of in vivo pharmacokinetics and metabolism to SARM activity and the inability of in vitro models to predict in vivo response. With this understanding, we screened a series of AR nonsteroidal ligands. We identified S-22 as a compound with the most potent and tissue-selective in vivo activity that we have observed to date and favorable pharmacokinetic properties. These compounds are promising for clinical use on androgen-deficiency related disorders. Our continuing studies will evaluate the preclinical and clinical value of identified SARMs and optimize the chemical structures based integrated structural, pharmacological, and pharmacokinetic data.