Novel Androgen Receptor Antagonists for the Treatment of Prostate Cancer - Accounts in Drug Discovery Case Studies in Medicinal Chemistry - page 127

(KO) mice, thus demonstrating that convulsion was not a necessary con-
sequence of the intended pharmacology. Electroencephalography in the rat
established that the convulsions were caused by classical seizure, and drug
concentration measurements in brain and plasma across a structurally diverse
set of AR antagonists established the generality of steep dose-responses for
convulsion and the importance of brain penetration. In vitro proconvulsive
target screening studies revealed that the common off-target pharmacology of
GABA A inhibition was shared by all AR antagonists tested (IC 50 ¼ 7-47 mM)
and spanning a broad range of chemical structures (e.g., including the AR
antagonists bicalutamide, 2-hydroxyflutamide, nilutamide, and the imide-based
BMS anti-androgens). These results led to the hypothesis that antagonism of
the GABA A receptor was causal for convulsions when combined with high
brain drug concentrations and successful drug competition for the steroid sites
occupied by endogenous GABA A modulators.
The outcome of these studies posed significant challenges for modifications in
the screening strategy. Efforts to identify potent AR antagonists that were
devoid of measurable GABA A antagonist activity were not successful. A
potential approach to reduce seizure risk would be to identify compounds with
limited distribution to the CNS. However, no high throughput assay was
available to reliably predict distribution of compounds to the CNS at high
doses. Therefore, compound selection had to rely on assessment of low
throughput in vivo properties determinative of safety margin, such as plasma
and brain drug concentrations corresponding to the observed CNS events, and
projected human C max for the ecacious dose. Therefore, the previously
described screening paradigm was altered to include assessment of CNS effects
in mouse single dose toxicokinetic and tolerability studies prior to evaluation of
compound activity in repeat-dose testing in the CWR-22-BMSLD1 tumor
xenograft mouse model. In these studies, clinical signs were continuously
monitored for four hours following administration of high doses (500-
1000mg kg 1 ) of the AR antagonists and drug concentrations in plasma and
brain were determined at four hours post-dose. Plasma and brain exposure data
from the IRPW study were also utilized to get an early read on the brain
penetration potential of the lead compounds.
6.3.3 Improved CNS Profile with BMS-305
Further optimization using the modified screening strategy identified BMS-305,
a close analogue of BMS-949 in which the trifluoromethyl group at the 3 0 -
position of the aniline ring was replaced by an iodide (Figure 6.4). As shown in
Table 6.1, this simple change provided a 4-fold increased potency in the in vitro
assays. BMS-305 was ecacious in the CWR-22-BMSLD1 tumor model when
administered orally at a low dose of 10mg kg 1 day 1 (C max ¼ 26 mM). 12
Despite achievement of very high brain exposure (293 mM), BMS-305 did not
produce significant CNS effects in mice upon daily oral administration of a
1000mg kg 1 dose. No convulsions were observed in exploratory rat and dog