`
`Donald J. P. Pinto, Joanne M. Smallheer, Daniel L. Cheney, Robert M. Knabb, and Ruth R. Wexler*
`
`Research and Development, Bristol-Myers Squibb Company, P.O. Box 5400, Princeton, New Jersey, 08543
`
`Received February 3, 2010
`
`1. Introduction
`
`Thrombosis is the underlying cause of a host of common,
`debilitating, and often fatal cardiovascular disorders. Forma-
`tion of thrombi in the arterial circulation can lead to acute
`myocardial infarction (MIa) or ischemic stroke. In the venous
`circulation, deep vein thrombosis (DVT) may result in chronic
`leg pain, swelling, and ulceration and can, if partially or fully
`dislodged, be followed by life-threatening pulmonary embolism
`(PE). The public health consequences of thromboembolic
`disease are vast. For example, approximately 2.5 million people
`in the United States are affected by atrial fibrillation (AF), a
`cardiac arrhythmia associated with a 4- to 5-fold increase in
`the risk of stroke of primarily cardioembolic origin.1 Another
`group at elevated risk of ischemic stroke, as well as recurrent
`acute MI, is the large and growing population of patients with
`acute coronary syndrome (ACS).2 Furthermore, it has been
`estimated that DVT and PE, which together comprise venous
`thromboembolism (VTE), afflict up to 600 000 individuals in
`the United States each year and are implicated in at least
`100 000 deaths.3
`Despite the continued morbidity and mortality caused
`by thromboembolic disease, recent advances in drug develop-
`ment provide cause for optimism that we are about to enter a
`new era in antithrombotic therapy. One of the most important
`advances has been the development, and recent introduction
`into clinical practice, of a new class of anticoagulants, the
`direct factor Xa (FXa) inhibitors.4,5 In this Perspective, we
`provide a detailed insight into the development of these
`important new agents, describe how structure-based design
`played a pivotal role in this process, and review the wealth of
`preclinical and clinical data that have emerged to date.
`Finally, we consider the issues that will determine the future
`impact of the direct FXa inhibitors on clinical practice. We
`
`*To whom correspondence should be addressed. Phone: (609) 818-
`5450. Fax: (609) 818-3331. E-mail: ruth.wexler@bms.com.
`aAbbreviations: ACS, acute coronary syndrome; AF, atrial fibrilla-
`tion; aPTT, activated partial thromboplastin time; ATIII, antithrombin
`III; AV, arteriovenous; DTI, direct thrombin inhibitor; DVT, deep vein
`thrombosis; ECAT, electric-current-induced arterial thrombosis; FVa,
`factor Va; FVIIa, factor VIIa; FVIII, factor FVIII; FVIIIa, factor
`FVIIIa; FIX, factor IX; FX, factor X; FXa, factor Xa ; FXIa, factor
`XIa; GPIIb/IIIa, glycoprotein IIb/IIIa; HLM, human liver micro-
`somes; hERG, human ether-a-go-go related gene; HTS, high-through-
`put screen; LMWH, low-molecular-weight heparin; MI, myocardial
`infarction; PD, pharmacodynamic; PDB, Protein Data Bank; PE,
`pulmonary embolism; PK, pharmacokinetic; PT, prothrombin time;
`SAR, structure-activity relationship; TAFI, thrombin-activatable fi-
`brinolysis inhibitor; TAP, tick anticoagulant peptide; TF, tissue factor;
`TG, thrombin generation; THR, total hip replacement; TKR, total knee
`replacement; UFH, unfractionated heparin; VKA, vitamin K antago-
`nist; VTE, venous thromboembolism; vWF, von Willebrand factor.
`
`begin by briefly highlighting the limitations of the current
`standards of care in antithrombotic therapy, reviewing some
`key concepts in hemostasis and thrombosis, and explaining
`the rationale for targeting FXa.
`1.1. Current Antithrombotic Therapy. Numerous clinical
`trials have confirmed the efficacy of traditional anticoagu-
`lants, including vitamin K antagonists (VKAs), unfraction-
`ated heparin (UFH), and low-molecular-weight heparins
`(LMWHs, fractionated heparin with reduced activity toward
`thrombin compared to UFH), in the prevention and treat-
`ment of a range of arterial and venous thromboembolic
`diseases.1,6-8 Despite the fact that these drugs are the current
`standard of care and despite their proven efficacy, these
`anticoagulants all possess significant limitations that restrict
`their usefulness in the clinic and have created the need for
`new therapies. Use of warfarin and other VKAs is especially
`problematic, even though these anticoagulants offer the
`convenience of oral administration.9 For example, warfarin
`is associated with numerous drug and food interactions, an
`unpredictable pharmacokinetic (PK) and pharmacodynamic
`(PD) profile, and considerable intra- and interpatient varia-
`bility in drug response.9 As a result, the appropriate ther-
`apeutic dose varies, necessitating monitoring and frequent
`dose adjustment. Monitoring of warfarin therapy is critical
`because of this variability and relatively narrow therapeutic
`index, which often leads to subtherapeutic anticoagulation
`and a higher risk of thromboembolism or to excessive anti-
`coagulation and an increased risk of bleeding.9 Furthermore,
`the delayed onset of action of warfarin means that in critical
`situations therapy must be initiated with a rapid-acting,
`parenteral anticoagulant. Urgent surgical or invasive proce-
`dures may also be complicated by the fact that the anti-
`coagulant effects of warfarin are retained for several days
`after discontinuation of treatment.
`Agents for short-term anticoagulation include UFH,
`LMWHs, the indirect FXa inhibitor fondaparinux, and
`the direct thrombin inhibitors (DTIs) argatroban, bivali-
`rudin, and hirudin. These anticoagulants all require paren-
`teral administration, which makes their use outside the
`hospital problematic and which can also be associated with
`injection-site hematomas. UFH and, to a lesser extent,
`LMWHs carry the risk of thrombocytopenia, and since they
`are produced from animal tissue, they are sometimes asso-
`ciated with serious side effects.10 In addition, UFH has an
`unpredictable PK profile and anticoagulant response that
`necessitate monitoring.5 The limitations of parenteral anti-
`coagulants, particularly the need for injection, mean that
`warfarin and other VKAs, which in many countries are still
`the only orally administered anticoagulants approved for
`
`r 2010 American Chemical Society
`
`Published on Web 05/26/2010
`
`pubs.acs.org/jmc
`
`J. Med. Chem. 2010, 53, 6243–6274 6243
`DOI: 10.1021/jm100146h
`
`BMS 2001
`CFAD v. BMS
`IPR2015-01723
`
`
`
`use, represent the sole viable option for long-term anti-
`coagulation therapy.
`As discussed below, hemostasis and thrombosis are inter-
`related processes, and it is not surprising that anticoagulant
`and antiplatelet agents may have hemorrhagic complica-
`tions. Dose selection, therefore, is primarily influenced by
`the maximum dose that can be safely administered rather
`than the dose providing greatest efficacy. Although this will
`remain true for new antithrombotic agents on the horizon,
`benefit-to-risk profiles do vary.11 These limitations result in
`the underutilization of existing therapies in the very patients
`who are at highest risk of serious thrombotic events. For
`example, use of antithrombotic drugs is often avoided in the
`very elderly, despite the fact that advanced age is a major risk
`factor for both arterial and venous thromboembolism.12,13
`Concerns over bleeding have also restricted long-term use of
`warfarin and other anticoagulants in combination with
`antiplatelet drugs in patients with ACS and AF.4
`All of these factors have been the driving force for the
`development of new anticoagulants, including the direct
`FXa inhibitors. For any new anticoagulant to be considered
`ideal, it would need to possess a predictable PK profile that
`allows fixed oral dosing without routine monitoring and to
`possess a relatively wide therapeutic index with low peak-
`to-trough plasma concentrations to provide high levels of
`efficacy and low rates of bleeding. Since the target is in the
`central compartment, the ideal PK profile also features a low
`volume of distribution (to reduce the potential for off-target
`liabilities) and low systemic clearance (to minimize drug-
`drug interactions). Other key features of the “ideal” new
`anticoagulant include a rapid onset of action and an ability
`to bind clot-bound coagulation factors.14
`1.2. Hemostasis and Thrombosis. Hemostasis is the phy-
`siologic process during which bleeding is antagonized, and
`possibly stopped, in order to minimize blood loss.15 Unlike
`hemostasis, which is a necessary physiologic response to
`bleeding, thrombosis is a pathological process involving an
`exaggerated hemostatic response and is often ascribed to the
`combined influence of a triad of causative factors, namely,
`prothrombotic vascular endothelial changes or injury, stasis
`of blood flow, and/or other causes of hypercoagulability.16
`Primary hemostasis begins immediately after endothelial
`damage and involves localized vasoconstriction and the
`activation and adhesion of platelets to form a soft aggre-
`gate plug. In contrast, secondary hemostasis comprises a
`complex series of reactions during which several trypsin-like
`serine proteases,
`including FXa, are formed from their
`respective proenzymes. Figure 1A depicts how the “coagula-
`tion cascade” has classically been viewed, with two parallel
`and largely independent pathways, the extrinsic and in-
`trinsic pathways, converging at the point of factor X (FX)
`activation.4 Upon the formation of FXa, the common path-
`way is initiated, leading to the activation of prothrombin to
`thrombin and the subsequent conversion of fibrinogen to
`fibrin (described in more detail in section 1.3). Polymerized
`fibrin strands, together with activated platelets, then form
`stable clots that seal the breach at the site of injury. While this
`classic view of the coagulation cascade aids understanding of
`some of the key processes involved in secondary hemostasis,
`recent advances have led to the development of a cell-based
`model that may describe the time course and physiology of
`thrombogenesis more accurately,
`including the roles of
`tissue-factor (TF)-bearing cells and platelets. In this model,
`shown schematically in Figure 1B, coagulation occurs in
`
`Figure 1. (A) Classic view of the coagulation cascade. Sites of
`action of traditional and new anticoagulants are also depicted
`(coagulation factors affected by VKAs are indicated by an asterisk).
`Adapted by permission from Macmillan Publishers Ltd.: Clinical
`Pharmacology & Therapeutics (http://www.nature.com/clpt/index.
`html) (Gross, P. L.; Weitz, J. I. New antithrombotic drugs. Clin.
`Pharmacol.Ther. 2009, 86, 139-146),4 Copyright 2009. (B) Cell-
`based model of coagulation. Adapted by permission from Wolters
`Kluwer Health/Lippincott, Williams & Wilkins: Journal of Cardi-
`ovascular Pharmacology (Hammw€ohner, M.; Goette, A. Will war-
`farin soon be passe? New approaches to stroke prevention in atrial
`fibrillation. Journal of Cardiovascular Pharmacology 2008, 52,
`18-27),17 Copyright 2008, and Arteriosclerosis, Thrombosis and
`Vascular Biology (Monroe, D. M.; Hoffman, M.; Roberts, H. R.
`Platelets and thrombin generation. Arterioscler., Thromb., Vasc.
`Biol. 2002, 22, 1381-1389),18 Copyright 2002.
`
`three overlapping phases (initiation, amplification, and
`propagation) and is critically dependent on the contribution
`of cell surface proteins such as TF.17,18 In the initiation
`
`6244 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
`
`Pinto et al.
`
`
`
`phase, TF-factor VIIa (FVIIa) complexes activate factor IX
`(FIX) and FX; subsequently formed FXa-factor Va (FVa)
`complexes then generate small amounts of thrombin. During
`the amplification phase, thrombin generated in the initiation
`phase activates platelets, leading to release of factor VIIIa
`(FVIIIa) from factor VIII (FVIII)-von Willebrand factor
`(vWF) complexes on the platelet surfaces and also genera-
`tion of FVa and factor XIa (FXIa). In the final propagation
`phase, factor XIa (FXIa) generates more FIXa; this FIXa,
`together with that generated in the amplification phase,
`activates FX, leading to the formation of numerous FXa-
`FVa complexes and a burst of thrombin generation.
`The inhibitory activity of anticoagulants can be assessed
`by a number of methods. Assays that employ purified co-
`agulation proteases and synthetic substrates are most com-
`monly used for the initial characterization of the affinity
`of synthetic inhibitors toward their target enzyme. Modifi-
`cations of these methods can also be used to measure the
`activity of inhibitors in blood samples as, for example, the
`anti-Xa assay that can be used with the LMWHs, fondapar-
`inux, or synthetic inhibitors. Anticoagulants can be further
`characterized using clotting assays (including prothrombin
`time (PT) and activated partial thromboplastin time (aPTT))
`that measure the time for formation of a fibrin clot after
`addition of an activator of coagulation to whole blood
`or plasma.19 Clotting assays can be used for either in
`vitro characterization of the potency of compounds or for
`ex vivo assessments in laboratory animals or in humans.
`In vitro potency is conventionally reported as the concentra-
`tion of inhibitor required to produce a doubling of the
`uninhibited clotting time (PT2 or aPTT2). Other useful
`laboratory methods include the thrombin generation (TG)
`assay, which measures time-dependent changes in thrombin
`concentration.20
`1.3. Targeting FXa. FXa plays a critical role in coagula-
`tion. Together with FVa and calcium ions on a phospholipid
`surface, FXa forms the prothrombinase complex, which is
`responsible for the conversion of prothrombin to thrombin,
`the final effector of coagulation. Regulation of thrombin
`generation is the primary physiologic function of FXa, and
`few other roles have been identified.21 FXa is therefore an
`attractive and potentially specific target for new anticoagu-
`lant agents. As will be discussed, experience with indirect
`inhibitors of FXa has helped to validate FXa inhibition as
`an effective and safe anticoagulant strategy. However, in-
`direct FXa inhibitors possess two significant limitations.
`First, these agents require parenteral administration; second,
`they rely on the activity of antithrombin and are therefore
`not able to inhibit FXa bound within the prothrombinase
`complex.22 Development of orally administered direct inhi-
`bitors of FXa that can effectively inhibit prothrombinase-
`associated and clot-bound FXa, and thereby offer poten-
`tially greater anticoagulant activity, is therefore a highly
`significant advance.
`Once formed via the actions of FXa, thrombin plays key
`roles in both coagulation and platelet activation. Within the
`coagulation cascade, thrombin directly cleaves fibrinopep-
`tides from fibrinogen, participates in positive feedback reac-
`tions via activation of FV and FVIII, promotes the cross-
`linking of fibrin through activation of factor XIII (FXIII),
`and renders fibrin resistant to fibrinolysis through activation
`of thrombin-activatable fibrinolysis inhibitor (TAFI).23 Oral
`anticoagulant drug discovery efforts initially focused on the
`development of small-molecule anticoagulants that target
`
`thrombin directly, the oral DTIs. Although the rationale for
`targeting thrombin is clear, there is some evidence to suggest
`that inhibition earlier in the coagulation cascade at the level
`of FXa may have greater antithrombotic potential.21
`Furthermore, preclinical studies suggest that FXa inhibitors
`may possess a wider therapeutic index than DTIs.24 There-
`fore, it is not surprising that the oral anticoagulant drug
`discovery efforts of many pharmaceutical companies ulti-
`mately focused aggressively on small-molecule, direct FXa
`inhibitors. Differences between the direct FXa inhibitors and
`DTIs, and the potential consequences of these differences for
`clinical practice, are discussed in more detail in section 5.
`Before we begin our review of the development of the
`direct FXa inhibitors, it is worth briefly considering some
`important molecular features of the target protein. FXa
`belongs to the family of trypsin-like serine proteases, the
`catalytic domain of which consists of two similar antiparallel
`β-barrel folds that together form the catalytic triad and
`substrate binding site.25 Schechter and Berger have devel-
`oped a useful nomenclature to describe the prototypical
`binding site of a serine protease that has been widely adopted
`and that we will use herein.26 Accordingly, each protein
`subsite, labeled Si, binds the corresponding substrate amino
`acid labeled Pi, with “i” increasing toward the substrate
`N-terminus. Similarly, the corresponding subsites and sub-
`strate amino acids to the left of the scissile amide bond
`0
`0
`in Figure 2 are designated as Si
`, respectively, in-
`and Pi
`creasing toward the substrate C-terminus. Substrate clea-
`0-P1 amide bond. As the discovery
`vage occurs at the P1
`of small-molecule protease inhibitors has advanced, this
`convention has been extended to denote drug substructures
`that bind in a manner similar to substrate amino acids.27
`Figure 2A depicts the serine protease subsites primarily
`responsible for the recognition and binding of substrate
`and druglike molecules. It is noteworthy that all reported
`small-molecule serine protease inhibitors for which structur-
`al data exist bind in the S1 and one or more of the remaining
`subsites.
`The FXa binding site is defined by the S1 and S4 subsites
`and surrounding residues (Figure 2B and Figure 2C). S1 is a
`deep, largely hydrophobic cleft at the bottom of which lies
`the Asp189 and Tyr228 side chains. S4 is a strongly hydro-
`phobic pocket defined principally by the side chains of
`Tyr99, Phe174, and Trp215. The most potent ligands re-
`ported in the literature invariably engage both sites. Other
`features include the catalytic triad consisting of His57,
`Asp102, and Ser195 and the β-strand region defined by the
`214-217 backbone.
`Selectivity is a significant issue in the development of
`factor Xa inhibitors, since, as discussed above, several
`trypsin-like serine proteases play key regulatory roles in the
`coagulation cascade, among them FVIIa, FIXa, FXa, FXIa,
`and thrombin. Trypsin itself is an important enzyme for
`digestion of proteins in the gastrointestinal tract. Although
`the consequences of trypsin inhibition in humans have not
`been well-studied, results in laboratory animals could be
`problematic in preclinical toxicity testing that is required of
`any new drug.30 Orally administered drugs can reach con-
`centrations many-fold higher within the GI tract compared
`to concentrations within blood; therefore, a high degree of
`selectivity for the target coagulation enzyme over trypsin
`is important. In addition, selectivity over trypsin may be
`considered as a surrogate determination for achieving selec-
`tivity over other members of the trypsin-like protease family.
`
`Perspective
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
`
`6245
`
`
`
`Figure 2. Serine protease and factor Xa structure and nomenclature. (A) Depiction of substrate/protease nomenclature based on the
`convention of Schechter and Berger.26 This figure is based on that of Leung et al.28 (B, C) Overview of the factor Xa binding site,29 with subsites
`and several residues important for ligand binding labeled.
`
`As will be discussed in section 3 in more detail, a differentiat-
`ing feature that can be exploited for selectivity among serine
`proteases is the nature of the S1 pocket, being smaller and
`lipophilic in some such as trypsin, or larger and more
`hydrophobic such as in FXa.
`
`2. Development of Direct FXa Inhibitors
`
`2.1. Precedence for FXa Inhibition as an Effective and Safe
`Anticoagulant Therapy. Proof of principle for the effectiveness
`of direct FXa inhibition was established in preclinical animal
`models of thrombosis with naturally occurring FXa inhibitors
`of the prothrombinase complex such as tick anticoagulant
`peptide (TAP)31 and antistasin.32,33 Both are highly potent
`inhibitors of FXa (Ki = 0.59 and 0.3-0.6 nM, respectively),
`with >50 000-fold selectivity for FXa over other related serine
`proteases.34,35 Compound 1 (Figure 3), a synthetic pentasac-
`charide, is selective for FXa but acts indirectly via binding
`to antithrombin and has demonstrated improved or similar
`clinical benefit over LMWHs in venous thrombotic indica-
`tions.36 Superiority in ACS patients with unstable angina/non-
`ST-segment elevation MI for reducing risk of death or recur-
`rent heart attack was also demonstrated.37,38 The safety and
`efficacy of 1 provided the first clinical proof of principle that
`targeting FXa would be an important advancement in the area
`of anticoagulation therapy.36
`More recently, Sanofi-Aventis advanced a hypermethy-
`lated derivative of 1 with a high affinity for antithrombin III
`(ATIII), in which the amino functional groups were replaced
`with hydroxyl or methoxy groups.39 This compound, idra-
`parinux (2a, Kd = 1 nM, Figure 3), interacts more strongly
`with ATIII than 1 (Kd = 50 nM) and in patients has a longer
`half-life (t1/2 ≈ 80 h), allowing for once-weekly dosing. The
`results of phase II/III trials with 2a were mixed, however, and
`
`Figure 3. Structures of the pentasaccharide indirect FXa inhibitors.
`
`did not demonstrate a clear advantage over 1.40 In a phase III
`trial, long-term treatment with 2a (once weekly) for the
`prevention of stroke and systemic embolism in patients with
`AF was noninferior to warfarin but caused more bleeding.41
`Development of 2a has since been discontinued. A second-
`generation synthetic pentasaccharide, idrabiotaparinux (2b,
`SSR126517E), is in late-stage clinical trials for treatment of
`VTE and for stroke prevention in patients with AF,4,42
`although the company recently announced discontinuation
`of development of the drug for the latter indication.43 Com-
`pound 2b incorporates a biotin moiety that enables the
`selective reversal of anticoagulant activity by intravenous
`(iv) administration of avidin, which binds the biotin group.
`
`6246 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
`
`Pinto et al.
`
`
`
`was rapidly distributed in the plasma and rapidly eliminated
`with a half-life of 1.5-2 h.55 In a phase II trial setting, 4
`significantly reduced prothrombin fragment 1 þ 2, a marker
`of thrombin generation, when compared with UFH and a
`glycoprotein (GPIIb/IIIa) antagonist, eptifibatide.56 Two
`phase II clinical trials for the management of ACS and
`patients with ACS undergoing percutaneous coronary inter-
`vention have also been completed and showed the potential
`for reduced ischemic events with bleeding rates similar to
`those of UFH plus eptifibatide.56,57 A third parenteral agent
`that was advanced to human clinical trials was fidexaban (5,
`ZK-807834, Berlex-Pfizer; Figure 4),58 a compound that
`contains two amidine groups and a polar carboxylic acid
`moiety.59 The dihydrochloride salt of 5 (ZK-807191) is a
`potent inhibitor of FXa (Ki = 0.10 nM) and exhibits nearly
`20 000-fold selectivity over thrombin and 2500-fold selectiv-
`ity over trypsin; it has been shown to be efficacious in several
`in vivo animal models of thrombosis.60-62 In human clinical
`trials, infusion of 5 was found to be well tolerated.63 Phase II
`trials were underway in unstable angina in 2001; however,
`no results from these studies were published.64 In addition to
`the above direct FXa parenteral compounds, a dual throm-
`bin/Xa inhibitor, tanogitran (6, BIBT 986, Boehringer In-
`gelheim, FXa Ki = 26 nM, thrombin Ki = 2.7 nM, Figure 4),
`was evaluated more recently in a phase II clinical trial
`involving a human model of endotoxin-induced coagula-
`tion.65 In this study, 6 prolonged plasma aPTT, reduced in
`vivo thrombin generation in a dose-dependent manner, and
`was safe and well tolerated. No further development has
`been reported.
`Clinical success of indirect FXa inhibitors such as 1 and
`the improved therapeutic index with the early direct FXa
`inhibitors such as the parenteral inhibitors described above
`fueled an intense effort to discover and develop safer and
`more effective oral FXa inhibitors. The discovery and ad-
`vancement of orally bioavailable, direct-acting FXa inhibi-
`tors that progressed to clinical studies, including a brief
`survey of some of the early inhibitors that led up to these
`agents, are now discussed.
`2.3. Approach to Oral FXa Inhibitors. 2.3.1. Transition
`State and Peptidomimetic Approach: Covalent Inhibitors.
`Early efforts to identify inhibitors of FXa stemmed from
`the prior discoveries of thrombin inhibitors that contained
`“serine traps”, such as aldehyde or ketothiazole moieties,
`which are capable of interacting covalently with the cata-
`lytic Ser195 hydroxyl group to mimic a tetrahedral transi-
`tion state in a reversible manner. Compounds 7 (FXa IC50 =
`15 nM),66 8 (FXa Ki = 0.13 nM),67 and 9 (FXa IC50 =
`0.83 nM)68 are examples of early FXa inhibitors that contain
`a “serine trap” (Figure 5). As with the thrombin inhibitors in
`this class, several of the first transition-state FXa inhibitors
`also contained arginine or constrained arginine P1 residues
`that would interact strongly with the acidic Asp189 S1 resi-
`due, flanked by aromatic residues designed to fit into the
`hydrophobic S1 pocket of FXa.44 In the design of these
`inhibitors, it also became apparent that basic substituents
`could be tolerated in both the S1 and S4 regions, with a
`basic P4 moiety interacting in a π-cation manner with the
`hydrophobic residues in the S4 subsite. These peptide-like
`inhibitors eventually evolved to incorporate heterocyclic
`amide-bond replacements,69 such as pyridone (10, FXa
`IC50 = 3 nM), ketopiperazine (11, FXa IC50 = 2 nM), and
`caprolactam (12, FXa IC50 = 3 nM), while maintaining
`good FXa binding affinity; however, oral bioavailability was
`
`Figure 4. Parenteral direct FXa inhibitors 3, 4, and 5 and the dual
`FXa/thrombin inhibitor 6.
`
`2.2. Early Prototype Direct FXa Inhibitors: Parenteral
`Agents. The shortcomings of the LMWHs (e.g., indirect
`activity, limited ability to inhibit fibrin-bound FXa, paren-
`teral use only) provided great impetus for the discovery of
`synthetic, small-molecule direct FXa inhibitors. The first
`generation of these were iv agents, and several small-mole-
`cule, nonpeptidic, direct inhibitors of FXa were advanced
`to phase II clinical trials as parenteral agents.44,45 DX-9065a
`(3, Daiichi Sankyo, Figure 4, FXa Ki = 41 nM, thrombin
`Ki > 2000 μM, trypsin Ki = 620 nM) was clearly one of the
`first potent and selective FXa inhibitors identified, with
`good clotting activity (activated partial thromboplastin time
`twice the control [aPTT2] = 0.97 μM, prothrombin time
`twice the control [PT2] = 0.52 μM).46,47 The compound was
`studied extensively in preclinical models; found to be effica-
`cious in animal models of thrombosis after iv, subcutaneous,
`and oral administration; and did not show prolonga-
`tion of bleeding time.48 Because of its very low human oral
`bioavailability (F = 2-3%), 3 was advanced clinically
`as a parenteral agent.49 The compound was well tolerated,
`with no increase in bleeding over the dose range. Human PK
`for 3 showed a long half-life (t1/2 > 20 h), low clearance
`(120 mL/min), and low plasma protein binding (60% bound).
`Plasma concentrations correlated well with PD markers.
`In a phase II study in patients with non-ST-elevation ACS,
`dose-related trends toward reductions in ischemic events
`with high-dose 3 compared with heparin were observed.50
`Safety parameters such as bleeding increased dose propor-
`tionally.
`Otamixaban (4, FXV-673, Sanofi-Aventis, Figure 4), a
`2,3-disubstituted β-aminoester derivative, is a potent, rever-
`sible FXa inhibitor (Ki = 0.5 nM) with good in vitro antico-
`agulant activity (aPTT2 = 0.41 μM, PT2 = 1.1 μM).51,52
`Like 3, compound 4 belongs to the benzamidine class
`of molecules, and its polarity precludes significant oral
`absorption. In vivo, 4 was efficacious in canine models of
`thrombosis and demonstrated minimal effect on bleeding at
`effective doses.53,54 In phase I/II studies, 4 was administered
`intravenously and was found to be well tolerated in healthy
`volunteers and patients with coronary artery disease and
`
`Perspective
`
`Journal of Medicinal Chemistry, 2010, Vol. 53, No. 17
`
`6247
`
`
`
`benzoxazinone ring resulted in analogues, such as 21, that
`retained potent binding affinity against FXa (Ki = 6 nM).
`Kissei Laboratories reported the discovery of 22 (KFA-
`1411, FXa Ki = 1.7 nM), which is highly selective over a
`broad range of serine proteases,
`including trypsin, and
`has good in vitro anticoagulant potency (PT2 = 0.28 μM,
`aPTT2 = 0.87 μM).78 The compound was efficacious when
`compared with dalteparin and LMWH in a hemodialysis
`model in monkeys.79 Additional early bisamidine-containing
`analogues have been covered extensively in previous review
`articles.80-82
`2.4. Transition From Benzamidine to Oral Agents. 2.4.1.
`Isoxazoline, Isoxazole, and Pyrazole-Based Inhibitors. In the
`quest for potent inhibitors of FXa, Bristol-Myers Squibb
`researchers recognized the similarity between the platelet
`GPIIb/IIIa peptide sequence Arg-Gly-Asp and the pro-
`thrombin substrate FXa sequence Glu-Gly-Arg.83 Isoxazo-
`line derivative 23 (FXa Ki ≈ 39 000 nM, Figure 7) was
`
`identified from a high-throughput screen (HTS) of a library
`of proprietary GPIIb/IIIa antagonists. Modifications to
`incorporate the bisbenzamidine motif of the known FXa
`inhibitors, along with further optimization, provided 24
`with greatly enhanced affinity (FXa Ki = 94 nM).83 Guided
`by structure-based design, these researchers replaced the
`4-amidino moiety of 24 with a neutral o-phenylsulfonamide
`group to reduce the basicity of this dibasic lead and thus
`improve permeability. This afforded the first known mono-
`basic FXa inhibitor 25, which also exhibited enhanced
`potency (FXa Ki = 6.3 nM).84 The biaryl moiety of 25 was
`designed to interact with the hydrophobic S4 aryl binding
`domain of the FXa active site (see Figure 2B) and, from
`modeling experiments, was shown to be neatly stacked
`between the residues Tyr99, Phe174, and Trp215, with the
`terminal o-phenylsulfonamide ring designed to make an edge-
`to-face interaction with Trp215. Further scaffold optimiza-
`tion led to isoxazoline compound 26 (FXa Ki = 0.55 nM)
`with the benzamidine P1 and the biarylsulfonamide vicinally
`substituted on the five-member ring.85 Replacement of
`the isoxazoline ring with a planar aromatic isoxazole ring
`in 27 provided further improvement in FXa potency (Ki =
`0.15 nM).85 The lack of chirality of this template, and the
`potency, made it an attractive starting point for further
`optimization. Rapid evaluation of a variety of vicinally
`substituted five- and six-member ring scaffolds resulted in
`the determination that the five-member nitrogen-based,
`N-linked templates were most potent.86 This effort led to
`the discovery of a very potent pyrazole analogue 28 (SN429,
`FXa Ki = 0.013 nM).87 An X-ray crystal structure of 28 in
`bovine trypsin showed the following interactions: S1-Asp189
`with the benzamidine P1, the pyrazole N2 group with the
`backbone of Gln192, a lipophilic interaction of the C3 methyl
`at the outer ridge of the enzyme, the linker carboxamide
`carbonyl oxygen with the NH of Gly216, and the P4 bi-
`arylsulfonamide neatly stacked in the S4 hydrophobic box
`formed by Tyr99, Trp215, and Phe174. In vivo, in the rabbit
`arteriovenous (AV) shunt model, administration of 28 by iv
`infusion resulted in reduction of thrombus weight by 50% at
`a dose of 0.02 μmol/kg/h (ID50).87 Poor oral bioavailability
`(F < 4% in dogs), combined with a short half-life (t1/2 =
`0.82 h) and lack of selectivity over related trypsin-like serine
`proteases, precluded further development of 28 as an oral
`anticoagulant.
`The high affinity of 28 for FXa afforded a unique oppor-
`tunity to give up potency in an effort to improve the oral
`
`Figure 5. Examples of transition-state inhibitors of FXa.
`
`not achieved. There are no published reports of any transi-
`tion-state FXa inhibitors advancing into clinical trials.
`2.3.2. Early Dibasic Benzamidine Approach. Reports of
`nonpeptidic, small-molecule inhibitors of FXa such as bis-
`amidine compounds 13 (DABE, bovine FXa Ki = 570 nM)70
`and 14 (BABCH, bovine FXa Ki = 610 nM)71 set the stage
`for the evolution of additional dibasic inhibitors (Figure 6),
`of which 3 is an early example. The success of 3 prompted
`multiple research groups to further optimize compounds in
`this class. Initial attempts at Daiichi Sankyo to improve FXa
`potency and oral bioavailability in direct analogues of 3
`resulted in constrained indoline compounds 15 (FXa IC50 =
`7.6 nM)72 and 16 (FXa IC50 = 3.9 nM).73 Although these
`analogues demonstrated potent binding affinity against
`FXa, selectivity over trypsin remained an issue (trypsin Ki =
`24 and 39 nM, respectively). This was addressed by mod-
`ification of the naphthyl P1 moiety to a substituted benz-
`amidine and introduction of a hydroxyl mo