Small molecule kinase inhibitors for the treatment of brain cancer

In addition to each of the factors that govern the identification of a successful oncology drug candidate, drug discovery aimed at treating neurological cancer must also consider the presence of the blood−brain barrier (BBB). The high level of expression of efflux transporters (e.g., Pglycoprotein (Pgp) and breast cancer resistance protein (Bcrp)) at the BBB limits many small molecules from freely reaching the brain, where neurooncologic malignancies reside. Furthermore, many of the targets identified for the potential treatment of central nervous system (CNS) malignancies suggest that kinase inhibitors, capable of penetrating the BBB to reach their target, would be desirable. This Perspective discusses the unmet need for neurooncology treatments, the appeal of kinase targets in this space, and a summary of what is known about free brain penetration of clinical inhibitors of kinases that are of interest for the treatment of brain cancer.

Small Molecule Kinase Inhibitors for the Treatment of Brain Cancer Timothy P Heffron *

Genentech, Inc., 1 DNA Way, South San Francisco, California 94080, United States

*S Supporting Information

ABSTRACT: In addition to each of the factors that govern the identi fication

of a successful oncology drug candidate, drug discovery aimed at treating

neurological cancer must also consider the presence of the blood −brain barrier

(BBB) The high level of expression of e fflux transporters (e.g., P-glycoprotein

(P-gp) and breast cancer resistance protein (Bcrp)) at the BBB limits many

small molecules from freely reaching the brain, where neurooncologic

malig-nancies reside Furthermore, many of the targets identi fied for the potential

treatment of central nervous system (CNS) malignancies suggest that kinase

inhibitors, capable of penetrating the BBB to reach their target, would be

desirable This Perspective discusses the unmet need for neurooncology

treatments, the appeal of kinase targets in this space, and a summary of what is

known about free brain penetration of clinical inhibitors of kinases that are of

interest for the treatment of brain cancer.

Neurooncology encompasses the study of tumors that originate

in the brain (e.g., glioblastoma multiforme (GBM)) as well as

brain metastases In 2015, it was anticipated that more than

21 000 new cases of malignant brain and central nervous system

(CNS) cancers would be diagnosed in the United States that

GBM which has an associated poor prognosis (3-year survival

rate 3 −5%).2

Despite the apparent unmet medical need, there

has been little progress in developing new treatments for GBM.

Most evaluations of chemotherapeutics in GBM have failed.

Currently, the alkylating agents temozolomide (approved 2005)

and the carmustine-based Gliadel wafer (approved 1996) are the

only chemotherapeutics that are FDA approved for the treatment

of newly diagnosed GBM Other neurological cancers have

similarly limited drug treatment options.

In addition to the need for more treatment options for

primary brain tumors, metastasis of tumors to the CNS occurs

from as many as 40% of peripheral tumors, with well over

100 000 cases per year.3When a kinase inhibitor is used for the

treatment of peripheral disease, such CNS metastasis is a risk

as a mechanism of emergent resistance if that inhibitor is not

freely CNS penetrant In this scenario, treatment of a tumor

with drug is e ffective until disease progression occurs in the

CNS, where drug concentrations are limited As an example of

the signi ficance of the challenge presented by resistance due to

CNS metastases, 14% of patients with HER2-positive breast

cancer treated with pertuzumab had first evidence of disease

progression due to CNS metastasis, evidently as a result of

the inability of pertuzumab to cross the blood −brain barrier

(BBB).4Unfortunately, as discussed below, this scenario is not

limited to HER2-positive disease treated with a therapeutic

antibody but also happens with numerous FDA approved

small molecule kinase inhibitors that do not penetrate the CNS.

For CNS metastases, prognosis is generally poor and chemotherapy is useful only in limited settings,5 furthering the unmet need for new chemotherapeutics for malignancy in the CNS.

While primary brain tumors and brain metastases are distinct disease manifestations and may require targeting di fferent drivers of disease, for the medicinal chemist, the approach to treating each of these indications requires the same consid- erations of the BBB, which typically limits small molecule penetration to the CNS where brain tumors reside Furthermore, for both primary and secondary brain tumors there is biological rationale to develop BBB penetrating kinase inhibitors While there have been 32 kinase inhibitors approved for the treatment of cancers that reside outside the CNS, no kinase inhibitor has been approved for the treatment of primary CNS tumors, while alectinib (61) has recently received accelerated approval to treat patients including those with brain metastases One reason for the lack of approved kinase inhibitors for treating brain tumors is that in order to e ffectively treat brain tumors, the kinase inhibitor must be capable of reaching its target Therefore, the kinase inhibitor must effectively cross the BBB As will be discussed below (and included as Supporting Information ), the majority of approved kinase inhibitors and kinase inhibitors that have advanced to clinical study have no report of CNS penetration, reportedly limited CNS pene- tration, or CNS penetration that is expected to be limited due

to the action of the e fflux transporters P-glycoprotein (P-gp) and breast cancer resistance protein (Bcrp).

When considering potential therapeutics for the treatment of brain cancer, it is frequently asserted that because of disruption

of the BBB by primary tumors or metastases in the brain,

consideration of the BBB is not relevant However, while it may

be true that a tumor can disrupt the BBB, it generally does

so just partially and significant literature reports indicate the

importance of the BBB in limiting drug penetration to its intended

target even when a tumor causes such partial disruption.6

Addi-tionally, GBM in particular is noted to grow in a di ffuse manner in

which a signi ficant portion of the tumor grows behind an intact

BBB, and so without e ffective drugs that are capable of freely

crossing that barrier the tumor progresses.7 That GBM grows

in such a manner so as to remain behind an intact BBB

punctu-ates the need for small molecules to be able to penetrate that

barrier if they are to have potential to e ffectively treat that disease.

With an understanding of the importance of free BBB

penetration for drugs targeting brain cancer, neurooncology

medicinal chemistry programs have much in common with

programs for other CNS diseases Fortunately, in recent years

there has been a much improved appreciation for the

require-ment to achieve su fficient free drug concentration in the brain,

if that is where the target resides A recent Perspective provides

an excellent review of the concepts of free brain penetration

that are essential to CNS and neurooncology programs alike

and pertinent to the remainder of the discussion within.8

Succinctly, it is important to note that it is critical that kinase

inhibitors that are intended to treat brain tumors achieve

therapeutically bene ficial free drug concentrations in the brain.

Indeed, a recent conference on CNS cancer drug discovery and

development emphasized the need for neurooncology

pro-grams to focus on achieving free brain penetration.9To assess

in preclinical studies whether e ffective therapeutic

concen-trations of a molecule cross the BBB, and therefore whether it

has a realistic chance of achieving e fficacy by the intended

mechanism, some assessment of free brain or, as a surrogate,

cerebral spinal fluid (CSF) concentrations is needed.8

To assess the extent to which a small molecule freely penetrates the BBB

(as opposed to just achieving a target free concentration in

the brain), a comparison of free brain or CSF concentrations to

free plasma concentrations is needed (Kp,uu) It is worth noting

here that in the discussion of free brain penetration of clinical

kinase inhibitors found below, the target therapeutic

concen-tration is not often available, and so an assessment of free

CNS penetration (Kp,uu or free brain-to-free plasma

concen-tration ratios), where available, is utilized for an assessment.

Where such values were available, we considered values of

<0.1 to be low and evidence of signi ficant limitation to free

BBB penetration, whereas values of >0.3 demonstrate a

signifi-cant degree of free CNS penetration.

In principle, therapeutic free concentration of drug might be

able to penetrate the BBB even with very low Kp,uu values.

In order for this to occur, however, there would need to be a

corresponding increase in systemic exposure that might

increase risk of unintended side e ffects To illustrate, a kinase

inhibitor with a Kp,uuof 0.1 would require 10 times the sytemic

exposure to achieve a therapeutic bene fit in the brain compared

to a kinase inhibitor equivalent in all aspects except for a Kp,uu

of 1.0 When targeting CNS disease, then, the importance of

maximizing Kp,uu is a signi ficant consideration and likely to

impact the safety/tolerability of a molecule at doses required to

achieve therapeutically bene ficial free concentrations in the

CNS.

With an understanding of the importance of achieving free

CNS penetration with molecules intended for the treatment of

brain cancers, a paramount requirement for achieving signi

fi-cant free drug concentrations behind the BBB is that the small

molecules are not strong substrates of P-gp or Bcrp, e fflux transporters highly expressed at the BBB.8,10Small molecules that are significant substrates of P-gp are anticipated to have limited free CNS penetration, and in the discussion of clinical kinase inhibitors below, molecules that are reported to be P-gp substrates are suggested to likely have limited CNS penetration For medicinal chemists interested in kinase inhibitors to treat brain cancer, avoidance of P-gp transport must be a focus so as

to maximize Kp,uu Considerations in the design of kinase itors (or any small molecule) to limit transporter mediated

inhib-e fflux include a number of physicochemical properties that can

be prospectively calculated Among the most critical properties

to consider are the reported correlations between topological polar surface area (TPSA) and/or the number of hydrogen bond donors (HBD) and the likelihood of P-gp mediated

e fflux.8ATP-competitive small molecule kinase inhibitors generally employ hydrogen bonding interactions with the hinge of the kinase, and oftentimes multiple hydrogen bond donors are utilized.11

As a result of the common use of frequent hydrogen bond donors within kinase inhibitors, overcoming the physicochem- ical property restraints that predict e fflux while maintaining other desirable attributes of kinase inhibitors, including potency, is a challenge Indeed, a comparison of the median values of physicochemical properties of 119 CNS approved

for clinical use (all indications) reveals signi ficant disparities ( Table 1 ) Whereas CNS drugs have a median value of 1 HBD,

approved kinase inhibitors have 2 Additionally, approved kinase inhibitors have a median TPSA value double that of approved CNS drugs Kinase inhibitors also tend to have signi fi- cantly higher MW and lipophilicity than CNS drugs.

For more than 30 years, kinase inhibitors have been the focus

of signi ficant pharmaceutical pursuit and the appeal of kinase inhibitors as potential therapeutics extends to the treatment of brain tumors and metastases.14 While the nature of kinase inhibitors, particularly ATP competitive versions, may have some constraints on physical properties to achieve potency that are contrary to what is typical for CNS drugs, realizing potent kinase inhibitors that are capable of signi ficant free brain pene- tration is possible However, free brain penetration has not been

a design consideration for many kinase inhibitor programs and in some cases may have been intentionally avoided.15

Even when intentionally seeking potent and freely BBB penetrant kinase inhibitors, there are, of course, limitations to available in vivo brain cancer disease models in which such

Table 1 Comparison of Median Values of Physicochemical Properties for Kinase Inhibitors Approved for Clinical Use and 119 Drugs Approved for CNS Indications

median property value

approved kinase inhibitors (n = 34) a CNS drugs

aKinase inhibitors approved for any indication through 2015.13

bMarketed CNS drugs Values obtained from ref12

10031

Table 2 Structures and Key Properties for VEGFR and PDGFR Inhibitors Advanced to Clinical Study for Brain Cancera

10032

gliomas ”16

As an example, the U87 model of glioblastoma is a

frequently studied GBM model used in orthotopic mouse

xenograft studies The use of the U87 model to assess whether

or not a molecule has potential in the treatment of brain cancer

is limited, however, as it is known to maintain a highly

dis-rupted BBB, not relevant to clinical disease, and not to grow in

the di ffuse manner observed in human patients in which the

tumor invades healthy brain with an intact BBB.17 For this

reason, the U87 and potentially other models may overestimate

the likelihood that an agent may provide therapeutic bene fit

in human GBM patients To understand whether the drug is

capable of reaching its target in brain tissue, an evaluation of

free brain-to-plasma ratios, free brain concentrations, change in

brain concentration between wild-type mice and transporter

knockout mice, or at least assessment of whether it is a

sub-strate of P-gp or Bcrp is desirable.

The basis for the interest in kinase inhibitors to treat brain

tumors begins with the underlying biology of CNS malignancy.

In the following sections, individual kinase targets with relevance

in CNS malignancy are introduced In many cases kinase

inhib-itors have been studied in clinical trials of patients with brain

tumors or metastases without success However, in many of

those cases limited CNS penetration of the kinase inhibitor may

have contributed to a lack of e fficacy We identify clinical kinase

inhibitors for the kinase targets, and within each section on a

given kinase target, available data related to brain penetration of

any clinical inhibitors of that target are summarized In the few

cases where BBB penetrating inhibitors of a kinase target for

brain cancer are reported, the medicinal chemistry e fforts leading

to this pro file are discussed Ultimately, we summarize whether

or not clinical brain penetrant inhibitors of kinase targets of

interest for neurooncology are available Finally, a comparison of

the physical properties of clinical CNS penetrant kinase

inhib-itors for brain cancer with those that have limited CNS

pene-tration reveals remarkable similarity, and disparity from

proper-ties of CNS drugs.

■ VEGFR AND PDGFR

Inhibition of angiogenesis has been established as a bene ficial

approach to treating cancer, and the potential of this approach

to cancer treatment extends to cancers in the CNS.18Targeting

vascular endothelial growth factor receptors (VEGFRs) has

been suggested to be of particular interest for the potential

treatment of neurological tumors, as it is a known driver of

angiogenesis in CNS tumors and found to be overexpressed in

this setting, particularly the highly vascularized GBM.19

Addi-tionally, platelet derived growth factor receptor (PDGFR), a

kinase frequently inhibited by VEGFR inhibitors, has been

identi fied as a potential target for the treatment of GBM due to

its high expression in this context.20Indeed, at least 14 itors of VEGFR and/or PDGFR have been evaluated for their potential in the treatment of CNS tumors ( Table 2 ), yet an unfortunate few would be expected to freely penetrate the BBB

inhib-to reach such tumors.

Cediranib (1, Table 2 )21and pazopanib (2, Table 2 )22have been studied in phase II and phase III trials in GBM patients but did not show a survival benefit.23 , 24The diffuse nature of GBM vasculature growth would require e ffective penetration of brain tissue by the inhibitors to maximize e fficacy However, both cediranib and pazopanib have been reported to be sub- strates of both P-gp and Bcrp in vitro and these transporters were found to limit brain exposure in mice.25,26

Like cediranib and pazopanib, sunitinib (3),27sorafenib (4),28nintedanib (5),29tivozanib (6),30and dovitinib (7)31were each ine ffective in clinical GBM studies.32 − 36

Sunitinib, sorafenib, and nintedanib each are likely to have limited CNS penetration, as they are substrates of P-gp and/or Bcrp, whereas data are not available for tivozanib or dovitinib.37−39 Furthermore, for sora- fenib, another study suggests that patients treated with renal cell carcinoma treated with sorafenib progress due to metastases only observed in the CNS, suggesting a sanctuary from drug due to lack of BBB penetration.40

Regorafenib (8, Table 2 ) demonstrated an e ffect in a rat model of glioblastoma41and, accordingly, advanced to clinical studies for the treatment of GBM.42 While results are not available, e fflux transport may limit free concentrations of regorafenib behind the BBB as the molecule is a substrate of P-gp and Bcrp and in P-gp/Bcrp knockout mice a 5.5-fold increase in brain concentration was achieved when compared to wild type mice at the same time point.43

The PDGFR- β, c-KIT, and Flt3 inhibitor tandutinib (9, Table 2 ) was found to be a substrate of both P-gp and Bcrp, which limits brain exposure in mice.44 Still, tandutinib was advanced to a phase I clinical trial in patients with GBM In that study, brain concentrations in 6 patients were determined, and a mean brain-to-plasma ratio (total) in these patients was determined

to be 0.33 However, no free brain-to-free plasma ratios or free brain concentration data from subsequent studies have been reported, and so no conclusion can be made about whether or not su fficient target engagment was achieved.45

After demonstrating in vivo efficacy in three different orthotopic GBM models in mice,46 axitinib (10, Table 2 )47encouragingly demonstrated activity in a phase II study of patients with GBM.48 However, it remains possible that the

where some tumor typically resides behind an intact BBB, may

be limited due to the fact that axitinib is a signi ficant substrate

of P-gp and Bcrp This was demonstrated in mouse cokinetic studies in which P-gp/Bcrp knockout mice had

pharma-Table 2 continued

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10033

14- and 21-fold increases in brain concentration at 1 and 4 h

postdose when compared to wild type mice.49

Vandetanib (11, Table 2 )50 and lenvatinib (12, Table 2 )51

are additional VEGFR inhibitors that have advanced to clinical

studies to treat GBM52,53despite being reported substrates of

P-gp.54,55The ability to achieve e fficacious free concentrations

in the brain is a concern as P-gp e fflux is anticipated to limit

drug penetration to portions of tumor where the BBB remains

intact.

Whether or not vatalanib (13, Table 2 )56is a substrate of

P-gp or Bcrp in vitro has not been reported, and there are not

reports of brain penetration of this molecule either preclinically

or clinically Vatalanib was studied in phase I clinical trials in

patients with glioma or GBM, but development of the molecule

was halted prior to complete assessment in this patient

pop-ulation.57

Cabozantinib (14, Table 2 )58and brivanib (15, Table 2 )59

stand out among the VEGFR inhibitors discussed here, as they

are reported to not be substrates of P-gp transport, suggestive

of their potential in the neurooncology setting.60,61To the best

of our knowledge, there are no reports of preclinical in vivo studies describing free brain exposure or clinical study results evaluating brivanib for the treatment of CNS tumors However, consistent with its lack of P-gp transport, cabozantinib has undergone a phase II study for the treatment of GBM and demonstrated some clinical and pharmacodynamic activity.62Among the 15 VEGFR/PDGFR inhibitors discussed here and included in Table 2 , just two have been reported to have minimal P-gp mediated e fflux, of importance when targeting CNS malignancy That two, cabozantinib and brivanib, are able

to minimize P-gp transport demonstrates that it is possible to achieve with still potent kinase inhibitors and enables assess- ment of the validity of the hypothesis that inhibiting their targets might be an e ffective treatment approach for GBM.

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10034

lung cancer (NSCLC) was followed by the approval of afatinib

(18, Table 3 )65and more recently osimertinib (19, Table 3 ).66

In addition to their use in the treatment of NSCLC, erlotinib

and ge fitinib have been evaluated for the treatment of NSCLC

brain metastases that harbor activating mutations of EGFR.

of EGFR mutant NSCLC brain metastases,67it is also reported

that such molecules are not as e ffective in the treatment of

brain metastases as peripheral metastases, suggesting limited

CNS penetration.68In this scenario, the inhibitor may be able

to e ffectively treat some, or a portion of, individual metastases

where the BBB is compromised, yet lesions behind the BBB

continue to grow Consistent with this theory, PET imaging of

11C-erlotinib showed accumulation of drug in a brain metastasis

but not in normal brain tissue These data suggest that where

the BBB is intact, a “sanctuary” for tumor remains.69

That erlotinib was not capable of freely crossing the BBB was also established in

a preclinical model of glioma.70Ge fitinib, afatinib, and osimertinib

have each also been reported to be substrates of both P-gp and

Bcrp, and so brain penetration of those EGFR inhibitors is

expected to be limited.71−73 Nevertheless, what free

concen-tration of afatinib that is capable of reaching CNS metastases

clinically.74

The interest in EGFR inhibitors for treating CNS cancer

extends beyond brain metastases in NSCLC to GBM

treat-ment In the most common and aggressive form of brain cancer,

GBM, overexpression of EGFR is encountered in approximately

40% of patients and half of these have an associated extracellular

mutation of EGFR (variant III).75 These factors suggest the

potential utility of a brain penetrant EGFR inhibitor Several

small molecule EGFR inhibitors, including ge fitinib and erlotinib,

have been approved for use in EGFR mutant NSCLC but,

despite clinical study, have not resulted in approval for the

treatment of gliomas.76

Investigation of brain penetrant inhibitors of EGFR would

therefore be of interest For rociletinib (20, Table 3 )77no

asso-ciated P-gp e fflux or brain penetration data have been reported.

However, recently there have been two reports of EGFR

inhib-itors that, while maintaining the quinazoline core of earlier

EGFR inhibitors, were reportedly designed to e ffectively

pene-trate the BBB to allow for e ffective treatment of CNS disease.

The first, NT113 (21, Table 3 ), a pan-ERBB inhibitor,

demon-strated e fficacy in intracranial GBM xeongrafts, including those

with high EGFR vIII expression.78 A limitation in the

char-acterization of 21 is that, while brain-to-plasma ratios are

reported, no free brain concentrations or free brain-to-free

plasma ratios are reported, limiting the interpretation of just

how e ffectively this molecule penetrates the BBB Nevertheless,

in intracranial GBM xenograft studies, 21 was more e fficacious

than either erlotinib or lapatinib, potentially indicating some

improved degree of e ffective CNS penetration.

A second recently disclosed quinazoline-based clinical

EGFR inhibitor intended to cross the BBB is AZD3759 (22,

Table 3 ).79The disclosure of 22 describes the directed e ffort

toward speci fically identifying a brain penetrating inhibitor of

EGFR for the treatment of CNS tumors, particularly CNS

metastases that arise in the course of treatment of EGFR

mutant NSCLC In order to achieve the excellent brain

pene-tration that 22 realizes compared to ge fitinib ( Figure 1 ),

improving physical properties to reduce transporter mediated

e fflux was emphasized in the optimization effort In this case,

the number of rotatable bonds had an apparent correlation with

positioned next to the NH of the aniline in 22 ( Figure 1 ) This positioning allows for intramolecular interaction of the F

associated with increased transporter mediated e fflux The structural modi fications relative to gefitinib did not have an apparent detrimental impact on potency, as 22 and ge fitinib are reported to have the same potency in a cellular assay employing

an L858R EGFR mutant cell line, suggesting its potential in the treatment of NSCLC with EGFR mutant positive brain metastases The team at AstraZeneca demonstrated the e ffective penetration of 22 across the BBB in preclinical species by reporting both Kpuu,brainand Kpuu,CSFvalues that show that the molecule achieves equivalent free concentrations on each side

of the barrier in rats The scientists at AstraZeneca went on to show extensive penetration of 22 into monkey brain in PET imaging studies 22 also demonstrated remarkable efficacy in an

in vivo model of brain metastasis In this model, 22 clearly differentiates itself from erlotinib, which was not efficacious when administered at the same dose level as 22.

While EGFR has been a long-standing target in GBM, previous molecules have not allowed for clinical conclusion on the validity

of the target as transporter mediated e fflux does not allow them

to freely penetrate the BBB to where tumors reside The recent emergence of 21 and, particularly, 22 highlights an exciting opportunity to study inhibition of a known driver of a signi ficant percentage of GBM cases and NSCLC brain metastases 21 and

22 are also part of a very limited set of kinase inhibitors reportedly speci fically designed for the treatment of brain cancer.

■ PI3K/AKT/mTOR

In addition to targeting EGFR directly, another approach to treat GBM would be to target downstream kinases The phos- phoinositide 3-kinase (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin (mTOR) kinases comprise one such path- way and are implicated in a signi ficant percentage of GBM and

pathway is also suggested as a mechanism to treat human epidermal growth factor receptor 2 (HER2)-positive brain metastases.84As a result of this biological implication and the pursuit of inhibitors of this pathway for other tumors, a number

of agents have advanced to clinical trials in GBM patients.85

As the inhibitors of the PI3K/AKT/mTOR pathway have been reviewed in this context previously,85we provide here a brief summary organized according to primary target of the inhibitor.

Of the many PI3K/mTOR inhibitors that have entered clinical study, GDC-0084 (25),86 buparlisib (26),87 PX-866 (28),88

Figure 1 Structural modifications upon gefitinib (17), focused onreducing rotatable bonds and effective hydrogen bond donors, led tothe freely BBB penetrating inhibitor of EGFR, 22

10035

pilaralisib (29),89 and XL765 (30)90 have been part of trials

speci fically for GBM ( Table 4 ).83Buparlisib has also advanced

to clinical studies for the treatment of breast cancer patients

with brain metastases.91However, among these, only 25 was

apparently designed to ensure signi ficant free brain penetration.

In order to realize 25, a program was initiated to purposefully

identify a PI3K/mTOR inhibitor capable of crossing the BBB

so that it would be amenable to treating GBM These studies

began with GNE-493 (23)92 as a starting point which was a

potent inhibitor of PI3K and mTOR but was a substrate of

P-gp and Bcrp ( Figure 2 ).86In order to realize brain penetrant

analogs, the importance of reducing the number of hydrogen

bond donors in 23 was identi fied as critical To further predict the likelihood of P-gp and Bcrp mediated e fflux, as well as metabolic stability, in silico evaluations were used to prospec- tively evaluate designs From these e fforts GNE-317 (24, Figure 2 ) was first identified which demonstrated that a brain penentrant PI3K inhibitor di fferentiated from a PI3K inhibitor that does not penetrate the BBB (2-(1H-indazol-4-yl)-6-(4-methanesulfonyl- piperazin-1-ylmethyl)-4-morpholin-4-ylthieno[3,2-d]pyrimidine (GDC-0941),93not shown) in that it had a PD e ffect in normal brain tissue and had improved e fficacy in in vivo brain tumor models.9424 was found to have unacceptable projected human clearance and so was further optimized to 25 ( Figure 2 ), a molecule

Table 4 Structures and Key Properties for PI3K Inhibitors Advanced to Clinical Study for Brain Cancer or FDA Approveda

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

Figure 2.Modifications of PI3K/mTOR inhibitors that resulted in the discovery of 25, a brain penetrating inhibitor with desirable metabolicstability

10036

that is of comparable potency and has similar ability to cross

the BBB to 24 but was projected to have more desirable human

pharmacokinetic properties The ability of 25 to potently inhibit

PI3K/mTOR signaling in the brain, along with its desirable

projected human pharmacokinetic pro file, led to its advancement

to clinical trials for the treatment of GBM.

The report of the discovery of buparlisib (26) does not indicate

that achieving brain penetration was a design consideration.87

However, in subsequent reports buparlisib has been reported to

e ffectively cross the BBB and inhibit PI3K pathway signaling in

preclinical95 and early clinical studies in patients with recurrent

GBM.96Unfortunately and despite inhibition of PI3K signaling in

patient tumors, there was not substantial e fficacy Additionally,

buparlisib has been reported to cause mood changes, a side e ffect

not observed with other PI3K inhibitors in the clinical setting.84

The structurally related dual PI3K/mTOR inhibitor PQR309

(27) is reported to not be a substrate of P-gp97 and achieves

equivalent brain and plasma concentrations,98 although free

concentrations were not reported.

The inhibitors 28, pilaralisib (29), and 30 have each progressed to clinical trials for the treatment of GBM, but there has not been a report of whether brain penetration was a design consideration or if these molecules penetrate the BBB The PI3K and PI3K/mTOR inhibitors discussed above inhibit each of the class I PI3K isoforms ( α, β, δ, and γ) However, the only as yet approved PI3K inhibitor is idelalisib (31), a selective inhibitor of the δ isoform of PI3K.99

Idelalisib

is approved for the treatment of chronic lymphocytic leukemia.

We were unable to identify any indications that CNS tumor progression is a mechanism of resistance to idelalisib This is a potential risk, as idelalisib is reported to not penetrate the BBB,100 consistent with the disclosure that it is a substrate of both P-gp and Bcrp.101

Among mTOR inhibitors, the mTORC1 inhibitors olimus (32), temsirolimus (33), and sirolimus (34) are FDA approved agents ( Table 5 ).102Each of these molecules has been studied in patients with GBM but has not provided bene fit.83Perhaps insu fficient brain penetration is a contributing factor to Table 5 Structures and Key Properties for mTOR Inhibitors Advanced to Clinical Study for Brain Cancera

ever-aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10037

the lack of e fficacy, as everolimus and sirolimus are reported

to be substrates of P-gp (and temsirolimus is a prodrug of

sirolimus).103

inhibitors that have advanced to GBM clinical trials.106,107For

35, there is no indication of whether the molecule penetrates

the BBB.108In a clinical study of 36, GBM tumor-to-plasma

ratios ranged from 16% to 77%.107However, it is not possible

to ascertain if su fficient concentrations to expect efficacy are

achieved as free concentrations were not reported, including

where the BBB is intact.

Most encouraging of the clinical mTOR inhibitors from the

perspective of trying to treat brain cancer, palomid 529 (37)

has been reported to e ffectively cross the BBB as brain

concen-trations were similar in pharmacokinetic experiments

compar-ing wild type mice and P-gp knockout mice.109This makes 37

one of a small set of clinical kinase inhibitors (the only apparent

mTOR inhibitor) where limited brain penetration would not be

a principal factor in limiting conclusion on the value of a target.

In addition to the inhibition of PI3K and mTOR, inhibition

of AKT has received signi ficant attention in this pathway.110

Among the clinical AKT inhibitors perifosine (38),111

8-(4-(1-aminocyclobutyl)phenyl)-9-phenyl[1,2,4]triazolo[3,4-f

][1,6]-naphthyridin-3(2H)-one (MK-2206, 125),112PBI-05204 (39),113

4-(2-(4-amino-1,2,5-oxadiazol-3-yl)-1-ethyl-7-{[(3S)-3-

piperidinylmethyl]oxy}-1H-imidazo[4,5-c]pyridin-4-yl)-2-methyl-3-butyn-2-ol (GSK690693),114 uprosertib,115 XL-418 (structure

not disclosed),

(S)-2-(4-chlorophenyl)-1-(4-((5R,7R)-7-hydroxy-

3-(3-(4-(1-aminocyclobutyl)phenyl)-5-phenyl-3H-imidazo[4,5-b]-pyridin-2-yl)pyridin-2-amine (ARQ-092),117we were only able to

identify some indication of the likelihood of brain penetration,

or advancement to a clinical study for use in brain cancer, for

perifosine and 39 ( Table 6 ) Ultimately, the allosteric AKT

inhib-itor perifosine was part of a trial in GBM patients but did not

demonstrate e fficacy,83

consistent with limited (total) brain tration preclinically.118In a preclinical study of 39, signi ficant total

pene-brain concentrations were achieved in rats but no assessment of

free concentration was determined.119Additionally, for 125, a trial

regarding the ability of the drug to pass through the blood −brain

barrier ”.83

Achieving brain penetration was not an apparent design

consideration for any clinical AKT inhibitor, and no clinical

AKT inhibitor is conclusively capable of achieving sign ficant

free brain concentrations, suggesting opportunity remains for AKT inhibitors that might be used to treat brain cancers.

Fibroblast growth factor receptor (FGFR) kinase has been suggested as a potential target for the treatment of brain cancer,120,121 and numerous FGFR inhibitors have entered clinical development.122 As many of the FGFR inhibitors are nonselective, with many inhibiting VEGFR and PDGFR (discussed above), the focus of this section is limited to selective FGFR inhibitors that have entered clinical develop- ment ( Table 7 ).123Those inhibitors include AZD4547 (40),124

in figratinib is capable of penetrating the BBB Similarly, we were unable to identify data informing the potential of erda fitinib, 40, or 43 to freely cross the BBB On the other hand and of interest for its potential for the treatment of brain cancer, ARQ 087 was reported to achieve free brain-to-free plasma concentration ratios of about 0.1 in rats.129

■ IGF-1R

Type I insulin growth factor receptor (IGF-1R) has been identified as a potential target for the treatment of brain cancers,120,130and numerous IGF-1R inhibitors have advanced

to clinical trials.131Among the clinical IGF-1R inhibitors ( Table 8 ) linsitinib (44),132 BMS-754807 (45),133 BVP-51004 (46),134XL-228 (47),135 and INSM-18 (48),136 there is no indication that achieving CNS penetration was a design consideration.

45 was demonstrated to have limited total brain penetration

in mouse studies,137 and 48 is believed to be a substrate of

transport or brain penetration of the other IGF-1R inhibitors Unfortunately, the available data suggest that no clinical IGF-1R inhibitors are suitable to evaluate whether inhibition of this target would be beneficial for brain cancer treatment.

Table 6 Structures and Key Properties for Select AKT Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10038

palbociclib (49, Table 9 )141 is approved for the treatment of

hormone-receptor positive breast cancer Regarding its

potential for the treatment of brain cancer, palbociclib was

demonstrated to provide a survival bene fit in a genetic mouse

model of brainstem glioma.142 Preclinical studies in three

intracranial mouse models of GBM showed that palbociclib

was e fficacious either as a single agent or in combination with

radiation.143However, palbociclib was found at 25-fold higher

concentration in tumor than in normal brain tissue, suggesting

that the molecule has limited penetration into the brain and the

BBB is compromised at the core of the tumor but not in normal

brain tissue Therefore, despite the reports of e fficacy in brain

cancer models, palbociclib may have free brain concentrations

is intact This would be consistent with being a substrate of P-gp,144 and ultimately an assessment of free brain concen- trations or Kpuu,brainis necessary to draw a conclusion about the merits of palbociclib for use in brain cancer.

Like palbociclib, abemaciclib (50)145 is reported to be a substrate of both P-gp and Bcrp.146However, in mice and rats,

Kpuu,brainis measurable at at least 0.2 and, while perhaps a model

of modest utility, abemaciclib demonstrated e fficacy in an orthotopic U87 GBM model in rats.146,147 While there is potential to further increase free brain penetration, given that some free brain exposure is attained with abemaciclib, it is encouraging that a trial studying abemaciciblib in breast cancer,

Table 8 Structures and Key Properties for IGF-1R Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

Table 7 Structures and Key Properties for FGFR Inhibitors Advanced to Clinical Studya

compd primary kinase target HBD TPSA (A 2 ) cLogP MW preclinical assessment of brain penetration

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10039

non-small-cell lung cancer, or melanoma patients with brain

metastases is currently enrolling.148 To the best of our

knowl-edge, there is no evidence of brain penetration for other CDK4/6

inhibitors, including the clinical inhibitor ribociclib (51).149

In addition to the potential of CDK4/6 inhibition in the

treat-ment of brain cancer, inhibition of CDK1 and CDK2 has been

sug-gested as having potential application in the treatment of GBM.150

The CDK inhibitors that have advanced to the clinic that inhibit

CDK1/2 tend to be broad spectrum CDK inhibitors ( Table 10 ).149

Flavopiridol (52, Table 10 )151 has been shown to inhibit GBM tumor models in mice, including in the intracranial setting.152 However, the value of the in vivo e fficacy may be limited by a compromised BBB, as flavopiridol is a substrate of both P-gp and Bcrp and signi ficantly increased brain exposure

is observed in P-gp/Bcrp knockout mice than in wild type mice.153

Seliciclib (53, Table 10 ) was reported to achieve a plasma ratio (AUC) of 0.3 after a 25 mg/kg oral dose to rats.154

brain-to-Table 10 Structures and Key Properties for CDK1/2 Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

Table 9 Structures and Key Properties for CDK4/6 Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10040

However, this only considers the total concentrations, and the free

concentration ratio is likely to be lower as seliciclib is a reported

substrate of P-gp.155

Dinaciclib (54, Table 10 )156 has demonstrated potential

utility in in vitro studies to treat GBM;157 however, we were

unable to identify any data suggesting whether or not this

mole-cule is likely to penetrate the BBB.

Despite inhibition of neuroblastoma tumor cell growth in the

in vitro setting,158 SNS-032 (55, Table 10 ) is a substrate of

P-gp and brain penetration is limited in wild type mice compared

to P-gp knockout mice.159

AT7519 (56, Table 10 )160appears to be a substrate of P-gp

as it has less of an e ffect on P-gp overexpressing cell lines.161

We were not able to determine if R547 (57)162and AZD5438

(58)163 are capable of crossing the BBB or are substrates of

e fflux transporters Furthermore, we were unable to identify

any indication that achieving CNS penetration was a design

consideration in the discovery of any of the inhibitors of the

CDKs 1 and 2 Taken together, there remains an apparent lack

of a CNS penetrating CDK1/2 inhibitor if clinical assessment

of this target for the treatment of GBM is to be assessed.

Outside primary brain tumors, there is a growing recognition of

the ability of peripheral tumors to escape therapy via metastasis

to the brain As a result, there are emerging best-in-class opportunities for kinase inhibitors that prevent resistance in that manner An example of this phenomenon is seen in the case of anaplastic lymphoma kinase (ALK) fusion positive NSCLC.164 Crizotinib (59)165was the first ALK inhibitor to show remarkable benefit to ALK-positive NSCLC patients.166Unfortunately, disease progression on crizotinib therapy is inevitable In fact, in the phase I and II trials of crizotinib in ALK-positive NSCLC, the most common mechanism of pro- gression on therapy has been reported to be through CNS metastases.167,166 Crizotinib has poor free brain penetration, consistent with signi ficant P-gp mediated efflux in vitro, which likely allows for metastases to find “sanctuary” in the CNS allowing for disease progression.168,169 As a result of relapse through CNS metastases, as well as resistance mechanisms including kinase domain mutations, there has been a substantial

e ffort in identifying next-generation ALK inhibitors, which have been reviewed elsewhere.170

Among the many next-generation inhibitors ( Table 11 ), several have been reported to be potentially e ffective in con- trolling CNS disease in ALK-positive NSCLC In a small set

of patients in a phase I trial, ceritinib (60)171was reported to achieve responses in patients with brain metastases, including patients who had progressed on crizotinib.172 For ceritinib, Table 11 Structures and Key Properties for ALK Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10041

medicinal chemistry optimization did not apparently focus on

achieving free brain exposure.171

While, to the best of our knowledge, P-gp transport or free

brain penetration was not apparently a factor in the design of

the molecule, alectinib (61)173 is reported to be e fficacious

in preclinical models of brain metastases and, aiding in the

positive interpretation of the e fficacy results, alectinib is

report-edly not a substrate of P-gp.174Most signi ficantly, alectinib has

been reported to achieve responses in patients with CNS

disease that did not respond to crizotinib, indicating some

degree of brain penetration.175

Entrectinib (62)176is reportedly capable of crossing the BBB

in preclinical species, although free brain concentrations were

not reported Encouragingly, a clinical report of a NSCLC

patient with brain metastases who had a response to entrectinib

suggests that the molecule may achieve signi ficant free CNS

penetration.177

ASP3026 (63)178was reported to achieve a brain-to-plasma

ratio (AUC0−24h) of 0.72 in mice.179Free brain-to-plasma ratios

were not reported, however Brigatinib (64)180is another

next-generation ALK inhibitor that has some preliminary indication

of e fficacy in a cohort of 10 crizotinib resistant patients with

brain metastases.181Brigatinib was able to achieve a response

in a mouse intracranial tumor model, and 11C brigatinib was

visualized within the tumor However, in these studies, the

extent to which the BBB was compromised was not reported

and higher concentrations of brigatinib were observed in

the intracranial tumor than in normal brain tissue.181 While

insu fficient preclinical data exist to understand the extent to

which free brigatinib is capable of penetrating the BBB, very

encouraging clinical data are emerging showing that brigatinib

is e ffective in treating ALK-positive brain metastases, where

crizotinib was ine ffective, suggesting meaningful CNS

pene-tration of brigatinib in these patients.182

X-376 (65)183and X-396 (structure not disclosed, 126) are

additional ALK inhibitors 126 is reported to have a

brain-to-plasma ratio in mice comparable to that of crizotinib (a P-gp

substrate with low CNS exposure in humans) However, the

investigators suggest that due to the greater potency of 126,

there may be potential for e fficacy in the CNS setting.183

No reports of free brain levels are reported, however.

While each of the aforementioned next-generation ALK

inhibitors ( Table 10 ) may hold potential for treating

ALK-positive malignancy, there is insu fficient preclinical data to

demonstrate that these molecules lack transporter mediated

e fflux and/or achieve significant free brain exposure.

PF-06463922 (66, Table 11 )169 clearly stands out among

the next generation ALK inhibitors, as achieving free brain

penetration was an evident design consideration 66, which has

activity against both EML4-ALK fusion proteins and crizotinib resistant ALK mutants, was designed to be a brain penetrant ALK inhibitor by reducing P-gp transport.169

As resistance to crizotinib is acquired through various kinase mutations as well as CNS metastases, the P fizer program sought to achieve a molecule that would simultaneously

analogs capable of potently inhibiting crizotinib resistant ALK mutations did not achieve a desirable balance of potency with physicochemical properties that allowed for both good metabolic stability and low P-gp mediated efflux Crystallo- graphic information was utilized to design subsequent ALK inhibitors with improved lipophilic e fficiency The “U-shape” that ligands adopted in a cocrystal structure with ALK inspired macrocyclic analogs A 12-membered ring lactam was found to improve human liver microsomal stability and reduce P-gp

Figure 3 ) Rather than focusing on the in fluence of single physicochemical properties on P-gp mediated e fflux, the Pfizer group emphasizes the need for multiparameter optimization due to an interplay of the in fluence of numerous physico- chemical properties on e fflux In particular, HBD, log D, and

MW were each considered in parallel In these studies, acyclic analogs had higher levels of P-gp mediated e fflux than the macrocycles with comparable MW, log D, and HBD 66, like the other macrocycles as well as acyclic analogs described, takes advantage of an intramolecular hydrogen bond between the aminopyridine and the adjacent ether oxygen This intramolecular hydrogen bond may e ffectively mask a HBD The authors also suggest that the macrocycles may have reduced e fflux relative to acyclic analogs due to a reduced number of rotatable bonds and

a 10% smaller solvent-accessible surface area.169Ultimately, 66 achieves a free brain-to-free plasma AUC ratio of 0.2 after oral administration in rats.

In summary, ALK is one of a few kinase targets, along with PI3K, EGFR, and PLK (below), where discovery programs have been reportedly directed speci fically at achieving brain penetrating inhibitors With such an ALK inhibitor undergoing clinical study currently, evaluation of the clinical hypothesis can take place.

A parallel to the resistance to initial ALK inhibitors via brain metastasis is observed in the treatment of HER2-positive breast cancer with antibody therapeutics Among patients treated with trastuzamab (Herceptin), CNS metastases emerge in approx- imately 30% of patients.184Lapatinib (68, Table 12 ) is approved

as a small molecule HER2/EGFR inhibitor,185 but preclinical studies showed that therapeutic concentrations were not achieved

Figure 3.Macrocyclization led to improved metabolic stability, potency, and reduced P-gp mediated efflux in this series of ALK inhibitors

10042

in brain metastases,186 suggesting that lapatinib does not

e fficacious in that setting Furthermore, a study of

brain-to-plasma concentration ratios in mice shows that lapatinib is a

substrate of P-gp and Bcrp.187 Indeed, in the clinical setting lapatinib was found to have variable and limited penetra- tion into HER2-positive brain metastases.188,91 Aside from lapatinib, at least a dozen additional small molecule inhibitors

Table 12 Structures and Key Properties for HER2 Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10043

of HER2 have been reported to have advanced to clinical study

( Table 12 ).189

Among the HER2 inhibitors that have advanced to clinical

studies, neratinib (69),190 a pan-HER inhibitor, and tucatinib

(70)191have advanced to trials for breast cancer patients with

brain metastases Unfortunately, a phase II trial of neratinib

in patients with HER2-positive brain metastases was not

successful The lack of e fficacy attained in the trial may be due

in part to the noted lack of CNS penetration of neratinib,

determined in preclinical studies192and consistent with a report

that it is a P-gp substrate.193For tucatinib, on the other hand,

an active metabolite is reported to achieve brain-to-plasma

concentration ratios ranging from 0.5 to 2.1 across several time

points after a 75 mg/kg oral dose in mice.194While only total

concentration ratios were reported, and so a true indication

of free brain penetration is not available, in other studies

tucatinib was capable of inhibiting p-HER2 in mouse brain

tissue, demonstrating some extent of free exposure

intracranial HER2-positive xenograft study.191

TAK-285 (71)195is a clinical inhibitor of HER2 and EGFR

that has been demonstrated to be capable of penetrating

the BBB in preclinical studies In rats, the free brain-to-free

plasma (AUC) ratio was 0.24 after a 75 mg/kg oral dose.196

Additionally, 71 was demonstrated to not be a substrate of

P-gp and to confer e fficacy in a mouse model of HER2-positive

brain metastases.197 Taken together, the data are supportive

of evaluating 71 in the clinical treatment of patients

HER2-positive brain metastases.

Dacomitinib (72)198inhibits epidermal growth factor receptors

and other tyrosine kinases A clinical study is looking at the safety

and e ffectiveness of using dacomitinib to treat HER2-positive

breast cancer patients with progressive brain metastases

(NCT02047747) Unfortunately, there are apparently no data

available on whether dacomitinib is a substrate of e fflux

trans-porters that are expressed at the BBB.

study of patients with GBM in which the tumor was surgically

resected after drug administration This study showed that 73

had concentrations in the tumor and brain that were greater

than in plasma, although no indication of free concentrations

has been reported that would help to interpret whether e fficacy

could have been expected.200

interest for the treatment of brain cancers In preclinical studies,

xenograft ( flank).202

While 74 was reported to confer a survival bene fit in an orthotopic model of glioblastoma in mice, we were

not able to identify reports of whether 74 is an e fflux

trans-porter substrate or if it is capable of free brain penetration.203

Unfortunately, a phase I trial of 74 enrolling GBM patients

needed to be discontinued due to toxicity of the molecule

without bene fit.204

Pelitinib (75)205is another HER2/EGFR inhibitor that has

advanced to clinical studies but is reported to be a substrate of

inhibitor CP-724,714 (76)207was reported to be a substrate of

both P-gp and Bcrp and, accordingly, would not be expected to

be able to effectively cross the BBB.208

CUDC-101 (77),209sapitinib (78),210and AST1306 (79)211

have advanced to clinical trials, but we were unable to identify

whether or not these molecules are substrates of e fflux transporters

or whether they are capable of free penetration of the BBB.

The data available for tucatinib and 71 demonstrate that brain penetrant inhibitors of HER2, which would be of evident interest in the treatment of HER2-positive brain metastases, are achievable, and it is encouraging that a potential treatment for HER2-positive brain metastases may be realized.

■ b-Raf/MEK

Historically, melanoma patients have a high frequency (>90%)

of brain metastasis development, and once present, these patients typically survive for less than 6 months.212Recently, inhibitors

dabrafenib (81)214) and mitogen activated protein kinase (MEK) inhibitors (cobimetinib (85)215and trametinib (86)216), along with the combination of dabrafenib and trametinib, have received approval for the treatment of melanoma As b-Raf (including V600E mutant) and MEK inhibitors are used in the treatment of melanoma, it is important to consider how

e ffectively these agents cross the BBB to either treat or prevent CNS metastases.217 Additionally, b-Raf218 and MEK219 have been identi fied as potential targets for primary brain tumors.

At least seven Raf inhibitors220and more than a dozen MEK inhibitors221have entered clinical studies.

Among the clinical Raf inhibitors ( Table 13 ), there have been con flicting reports of clinical response of CNS metastases

to vemurafenib (80)222or dabrafenib (81)223therapy, ranging from ine ffective at treating brain metastases to a case of a complete response Both vemurafenib and dabrafenib have been reported to be substrates of P-gp and Brcp and, furthermore, achieve little free brain penetration in mouse studies.224,225The varied reports and modest clinical response rates could be attributed to inconsistent disruption of the BBB, and potentially greater rates of response could be achieved if a b-Raf inhibitor were capable of free brain penetration Raf-265 (82)226 does not apparently e ffectively cross the

intracranially.227 We were unable to identify any information suggesting whether or not the clinical Raf inhibitors encora- fenib (83),228 XL281 (structure not disclosed), RO5212054 (structure not disclosed), ARQ-736 (84),229or its active meta- bolite, are capable of penetrating the BBB or are e fflux trans- porter substrates Together, the available data suggest that a BBB penetrating Raf inhibitor remains elusive, yet there would

be signi ficant potential value for such an inhibitor in treating cancers that metastasize to the brain.

The approved MEK inhibitors cobimetinib (85, Table 14 ) and trametinib (86, Table 14 ) have each also been reported to have limited free brain penetration in mice due to P-gp and Bcrp mediated e fflux.230 , 224

Additionally, neither cobimetinib nor trametinib inhibited pERK in normal brain tissue in mice after oral administration despite inhibition in the periphery, further illustrating a lack of CNS penetration by these molecules.231 Furthermore, when coadministered, dabrafenib and trametinib had low mouse brain exposure as well.224Among the other clinical MEK inhibitors,221we were unable

to identify any data that would suggest that GDC-0623 (87), binimetinib (88), selumetinib (89), CI-1040 (90), TAK-733 (91), RO5126766 (92), or WX-554 (structure not disclosed) would be likely to penetrate the BBB Nevertheless, selumatinib and binimetinib have been studied clinically for the treatment

of brain cancers.232,233

growth in a LN229 intracranial glioblastoma mouse xenograft

10044

tumor model study.219 Additionally, that 93 is capable of

penetrating the BBB in rats was demonstrated by signi ficant

inhibition of pERK in normal brain tissue after oral

admin-istration of the drug This molecule was found to have

neuro-logic toxicities clinically, and with the hope to avoid similar

toxicities, refametinib (94) and RO4987655 (95) were

re-portedly designed speci fically to not penetrate the BBB.234 , 235

Accordingly, despite substantial peripheral exposure in rat or

mouse studies, inhibition of pERK in brain tissue by those

reported to achieve minimal CNS penetration in rats.236

Pimasertib (97)221 has been reported to inhibit pERK in

mouse brain tissue, indicating some degree of CNS

pene-tration.237Consistent with those results, pimasertib was reported

to not be a substrate of e fflux transporters.238

E6201 (98),239an inhibitor of MEK and other kinases, has

been shown to achieve total brain-to-plasma concentration

ratios of 4.8 −6.4 in rodents, although free concentrations were

not reported To support the suggestion that some 98 is free to

engage its target in mouse brains, the molecule demonstrated a

survival benefit in a mouse model of brain metastases.240

While in no case was achieving free brain penetration of a

MEK inhibitor a reported design consideration, two clinical

MEK inhibitors discussed above have achieved signi ficant CNS

concentration in preclinical studies Worth noting, those two

MEK inhibitors (93 and 97) each have a dihydroxy

have four nominal hydrogen bond donors It is therefore

curious that the physical properties of 93 and 97 are so

inconsistent with the median values of marketed CNS drugs in

the HBD category, which is known to have a substantial impact

on P-gp e fflux However, both 93 and 97 have the potential for multiple intramolecular hydogen bonding interactions which

these molecules While the brain penetrant MEK inhibitors

o ffer opportunity to study their benefit in patients with brain cancer, 93 also highlights the additional risk that brain pene- tration of small molecule drugs adds to to an already di fficult development path Both on- and o ff-target activities in the brain have the potential to render a drug less tolerated, potentially limiting the ability of such a molecule to benefit patients with peripheral malignancy.

■ PLK1/AURORA KINASES

Both Polo-like kinase (PLK)241and Aurora kinases have been suggested as targets for the treatment of brain cancers.242,243Among the PLK1 inhibitors that have advanced to the clinic ( Table 15 ), volasertib (99),244BI 2536 (100),245GSK 461364 (101),246 rigosertib (102),247 and NMS-P937 (103)248 have been shown to be substrates of P-gp, and therefore, CNS penetration is expected to be limited.

The PLK1 inhibitor TAK-960 (104, Table 15 ), however, was reportedly selected among other analogs because it had reduced P-gp mediated e fflux, suggesting the potential for CNS pene- tration.249 In the discovery of 104, it was noted that incor- poration of a substituent capable of acting as a hydrogen bond acceptor adjacent to an amide, e ffectively eliminating a hydro- dgen bond donor via intramolecular hydrogen bond, was essential

to reduce P-gp mediated e fflux ( Figure 4 ).

At least 15 Aurora kinase inhibitors have advanced to clinical studies.250We were unable to identify considerations of achieving

Table 13 Structures and Key Properties for b-Raf Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10045

CNS exposure in the discovery of those molecules Among

those Aurora kinase inhibitors that have entered clinical studies,

any discussion of potential BBB penetration is limited to

tozasertib (105)250 and alisertib (106)250 ( Table 16 ) which

were each reported to inhibit tumor growth in orthotopic GBM

models in mice, which is insu fficient to understand the extent

to which those molecules are capable of freely penetrating the

BBB.243,251

Protein kinase C (PKC) has been identi fied as a potential target

for the treatment of GBM.252Among the small molecule PKC

inhibitors that have entered clinical studies for the treatment

of cancer ( Table 17 ) are bryostatin (107), enzastaurin (108),

midostaurin (109), and UCN-01 (110).253Among these, staurin was part of a trial for the treatment of GBM,254as well as for the treatment of brain metastases from lung cancer,255but in each case it did not demonstrate signi ficant benefit Along with enzastaurin, we were unable to identify any data suggesting whether or not bryostatin or midostaurin is capable of penetrating the BBB or is a substrate of e fflux transporters P-gp or Bcrp The PKC inhibitor 110 achieved total brain-to-plasma con- centration ratios of 0.5 −1.0 at five different time points after a 3.5 mg/kg dose to rats.256 However, free concentrations are not reported, and so a proper interpretation of free brain penetration is not available from this study Additionally, 110 is

enza-a reported substrenza-ate of P-gp, enza-anticipenza-ated to limit CNS etration.257

pen-Table 14 Structures and Key Properties for MEK Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

10046

■ ABL AND Src

The discovery of imatinib (111, Table 18 )258 has rightfully

been heralded as a remarkable success story in targeted

thera-peutics for the treatment of patients with Philadelphia

chro-mosome positive (Ph+) chronic myeloid leukemia (CML) or

acute lymphoblastic leukemia (ALL) Despite initial responses,

however, in addition to the emergence of resistance mutations,259

brain metastases lead to progression on imatinib therapy in nearly

20% of patients.260Such CNS resistance likely emerges due to

the poor ability of imatinib to cross the BBB, demonstrated

both preclinically and clinically, as it is a P-gp and Bcrp

sub-strate.261

Dasatinib (112, Table 18 ) was discovered as a next-generation

Bcr-Abl inhibitor that potently inhibits imatinib resistant mutant

forms of the enzyme.262Dasatinib also inhibits Src, among other

kinases,263which has been identi fied as a potential target for the

treatment of GBM as well as a target for patients with brain metastases.264 Like imatinib, dasatinib is a substrate of both P-gp and Bcrp which limits its brain penetration.265Yet, despite approximately 10% (total) brain penetration (relative to plama)

in an intracranial mouse model of Ph+ CML, dasatinib was able to demonstrate a therapeutic bene fit The authors attribute the e fficacy to the exceptional potency of dasatinib against the

Table 15 Structures and Key Properties for PLK Inhibitors Advanced to Clinical Studya

aProperties for 119 marketed CNS drugs are included for comparison.*Median value of 119 marketed CNS drugs.12

Figure 4.Reducing effective HBD count by intramolecular hydrogen

bonding reduced P-gp efflux among a set of PLK1 inhibitors

Table 16 Structures and Key Properties for Clinical Aurora Inhibitors for Which BBB Penetration Data Are Availablea

compd

primary kinase target HBD

TPSA (A 2 ) cLogP MW

preclinical assessment of brain penetration

105 pan-Aurora 3 102 4.8 465 insufficient data

106 Aurora A 2 105 6.2 519 insu fficient data CNS

drugs * N/A 1 45 2.8 305

aProperties for 119 marketed CNS drugs are included for comparison

*Median value of 119 marketed CNS drugs.12

10047