imaging in cns drug discovery...imaging agents able to quantify levels of misfolded proteins such as...
TRANSCRIPT
Imaging in CNS Drug Discovery
Roger N. Gunn, Ph.D.1,2,3,* & Eugenii A Rabiner1,4, FCPsych(SA)
1. Imanova Ltd, London, UK
2. Division of Brain Sciences, Imperial College London, Hammersmith Hospital Campus, London, UK
3. Department of Engineering Science, Institute of Biomedical Engineering, University of Oxford, Oxford, UK
4. Department of Neuroimaging, Institute of Psychiatry, Psychology and Neuroscience (IoPPN), King’s College London, London, UK
*Corresponding author
Professor Roger N Gunn,
Imanova Ltd, Burlington Danes Building,
Hammersmith Hospital, Du Cane Road,
London, W12 0NN, UK.
Email: [email protected].
Telephone: +44 (0)208 008 6000. Fax: +44 (0)208 008 6491.
Seminars in Nuclear Medicine
Updates in Molecular Brain Imaging
Vol 47, issue 1, Jan 2017
(20-‐25 double spaced pages with additional space for illustrations, tables, etc.)
Abstract
The discovery and development of CNS drugs is an extremely challenging process
requiring large resources, timelines and associated costs. The high risk of failure
leads to high levels of risk. Over the last couple of decades PET imaging has
become a central component of the CNS drug development process, enabling
decision making in Phase I studies, where early discharge of risk provides increased
confidence to progress a candidate to more costly later phase testing at the right
dose level or alternatively to kill a compound through failure to meet key criteria. The
so called “three pillars” of drug survival, namely; tissue exposure, target engagement
and pharmacological activity, are particularly well suited for evaluation by PET
imaging. This review introduces the process of CNS drug development before
considering how PET imaging of the “three pillars” has advanced to provide valuable
tools for decision-making on the critical path of CNS drug development. Finally, we
review the advances in PET science of biomarker development and analysis that
enable sophisticated drug development studies in man.
Introduction
CNS drug discovery and development is a long and difficult process with the delivery
of a successful new medicine taking around 10 years to complete from the initial
identification of the disease and target biology through to the launch of an approved
NME. In addition, a lot of resource is consumed through high levels of attrition, for
failed drugs that never overcome all the hurdles required to realise a viable CNS
drug. These hurdles include appropriate safety, PK, BBB penetration, target
engagement, pharmacological activity and efficacy. Thus, vast costs and long
timelines are associated with the approval of each NME – with over $2 billion
required to bring each NME to market, by current estimates [1].
The drug discovery and development process can be broken down into a number of
phases. Following the identification of the putative biological target the process
passes through a screen for compounds with the right properties (Figure 1.). This
leads to libraries of compounds being generated that are then subsequently tested
pre-clinically in vitro and in vivo, before a candidate compound is taken into human
testing. In humans Phase I (Safety & Tolerability), Phase IIA (Proof of concept &
dose ranging), Phase IIB (definitive dose finding), Phase III (Pivotal placebo
controlled trials and long term safety) precede compound registration and approval
by the relevant regulatory authorities. Post approval Phase IV studies are conducted
(Post marketing surveillance studies). The size and cost of these studies increases
throughout these phases, with costs increasing in proportion with increased sample
sizes and safety monitoring requirements.
Figure 1. Phases in the Drug Discovery and Development process.
These exponentially rising costs make good decisions early in the process
imperative to killing compounds that do not have the right characteristics for a
successful drug as early as possible. For example, performing a Phase III study with
a compound that does not sufficiently engage the biological target is an easy way to
waste the order of $100 Million.
Direct assay of human brain tissues are extremely challenging, and thus historical
approaches to CNS drug development relied heavily on peripheral pharmacokinetics
(PK) and clinical measures. These lack direct quantitative information on the levels
of brain exposure to the drug, target engagement and pharmacological activity – the
so-called three pillars of drug survival [2](Figure 2). These three pillars represent the
hurdles that a candidate molecule must overcome to become a successful centrally
acting drug. In a 2012 review of decision making for 44 of its drug programs in Phase
II, Pfizer revealed that, in 43% of decisions, it was not able to conclude whether
these three pillars had been met [2]. PET neuroimaging is unique in being able to
provide a direct quantification of parameters central to the three pillars in the human
brain in vivo, through the use of radiolabelled drugs or biomarkers [1, 3].
IND$Subm
i*ed
$
NDA
$Sub
mi*ed
$
FDA$Ap
proval$
Figure 2. The 3-Pillars of CNS Drug Survival that can be assayed with PET
neuroimaging: Tissue Exposure, Target Engagement and Pharmacological Activity.
The first CNS PET studies looking at measuring target engagement of a drug were
performed in the late 80’s by Farde et al [4] who used [11C]raclopride as a radiotracer
to explore the target-engagement of antipsychotics designed to interact with the
dopamine D2-receptor. Such studies were first directly incorporated into the drug
development process in the early 90’s [5]. This has led to PET target engagement
(or occupancy) studies of novel drug candidates becoming de-rigeur in the last two
decades, as big Pharma has adopted this technology to provide confidence in brain
penetration and rich dosing information from small human cohorts in Phase 1 studies
[1, 3, 6]. These studies deliver information on two of the three pillars in small cohorts
(n=6-12 subjects) in first time in human (FTIH) studies, providing an opportunity for a
very early go-no-go decision in the development process.
Clinical'Outcomes'
''''
Tissue&&Exposure&
Target&Engagement&
Pharmacological&Ac6vity&
Plasma'PK'''''
Biodistribution studies, are ones where the drug candidate itself is directly
radiolabelled, have also been performed [7-9]. These studies provide information on
whether the drug access across the blood brain barrier and its concentration in brain
tissue.
Measures of pharmacological activity along with stratification for trial entry with
imaging agents able to quantify levels of misfolded proteins such as Aβ and tau have
provided important readouts for Pharma in neurodegenerative diseases [10].
This review focuses on PET measures of these three CNS drug survival pillars,
Tissue Exposure, Target Engagement and Pharmacological Activity as well as the
methods of Biomarker development and Quantitative analysis that are critical for
their application. In particular we highlight the more recent evolution and
sophistication of these approaches that have led to their increased value. We signify
these recent innovations in sections identified with the “+” symbol.
Biodistribution
PET radiolabelling of a small molecule drug with either C11 and F18 allows for the
introduction of an imaging tag without altering the properties of the drug compound
itself. Subsequent intravenous injection of the labelled drug and careful quantitative
analysis of the dynamic PET data allows for the direct measurement of the drugs
concentration in brain tissue. Quantitative analysis involves the fitting of an
appropriate tracer kinetic model to the data that partitions the total signal between
radioactivity emanating from blood vessels and that from brain tissue[11]. The
resultant signal from brain tissue provides information on the delivery of the drug into
brain tissue ( 𝐾!) and the equilibrium partition coefficient of the free and non-
specifically bound drug, between brain tissue and plasma ( 𝑉!").
Biodistribution +
Classical biodistribution studies measure the total brain drug concentration.
However, a more important measure is that of the free concentration of drug in the
brain, as this is the portion that determines the level of binding to the biological
target, and this information coupled with an estimate of the in vitro affinity allows for
a prediction of the level of target engagement under the assumption that in vitro and
in vivo affinities are equivalent.
The desire to quantify the free drug concentration in the tissue, has led to the
combination of equilibrium equilibrium dialysis measures of the tissue non-specific
binding with PET measures in order to directly convert the PET measures into the
free drug concentration in brain [12, 13].
𝐶!" = 𝑓!" 𝑉!" 𝐶! (1)
where 𝐶!" is the free concentration of drug in brain tissue, 𝑓!" is the fraction of the
non-displaceable signal which is represented by the free drug, and 𝐶! is the
concentration of drug in plasma. Combining equation 1 with an estimate of the drug
target affinity enables estimation of target occupancy (𝑂𝑐𝑐),
𝑂𝑐𝑐 = !!"!!"! !!
(2)
where 𝐾! is the equilibrium dissociation rate constant for the drug.
These studies also allow for an assessment of whether there are any active transport
mechanisms for the drug, which would lead to concentration gradients across the
BBB. Under the assumption of passive diffusion the free drug concentrations in
tissue and plasma at equilibrium will be equal. Thus, combining estimates of the PET
equilibrium partition coefficient between tissue and plasma (in the absence of
specific binding) and equilibrium dialysis assays of the plasma (𝑓!) and tissue (𝑓!")
free fractions, then allows for an assessment of transport across the BBB,
𝑉!" = !!"!!
= !!!!"
!!" !!"
(3)
where 𝐶!" is the free concentration of drug in plasma. Rearranging, yields,
!!" !!"
= !!" !!"!!
(4)
with !!" !!"
~ 1 consistent with diffusion, !!" !!"
> 1 consistent with active influx and
!!" !!"
< 1 consistent with active efflux across the BBB.
Recent developments have extended this approach to studying macromolecules
where the much slower drug kinetics require the use of longer-lived isotopes such as
Zr89, the labelling of antibody fragments or pretargeting approaches [14-16].
Appropriate quantitative analysis of these data is still an active area of research with
requirements for cold doses of the drug to block peripheral uptake, internalisation of
the drug-target complex and loss of the label from the drug contributing additional
levels of complexity, not adequately accounted for by current models.
Target Engagement
Target engagement (or occupancy studies), which utilise a PET biomarker and
varying doses of the drug under investigation, provide some of the most valuable
decision making data that can be acquired as part of a FTIH Phase I study [17-24].
Occupancy studies are superior, and are to be preferred to biodistribution studies for
the following reasons. In addition to confirming BBB penetration these provide
information on the level of target engagement and easily facilitate the assessment
and comparison of back up molecules without further extensive radiochemistryif lead
compounds fail. Combined with information on the desired levels of target
engagement to achieve efficacy and/or drug exposure that will avoid unwanted
adverse effects, occupancy studies allow drug development teams to decide on
whether to progress a drug candidate into later phase efficacy studies of larger size,
complexity and cost. In addition, these studies allow the optimisation of the likely
therapeutic dose range for later phase efficacy studies, reducing the size and cost of
these.
Occupancy studies require an radioligand appropriate for the molecular target, and if
none exist already, may require radioligand development in parallel with the drug
discovery programme, so that the radioligand will be ready for use as part of the
FTIH study (successful development efforts can take 12-24 mths). By measuring the
target availability through the outcome measure binding potential (BPND) in both a
baseline and post-drug scan it is possible to calculate the fractional occupancy of the
target by the drug candidate,
𝑂𝑐𝑐 = !"!"!"#$%&'$ ! !"!"
!"#$
!"!"!"#$%&'$ (5)
By performing a number of such scans at different doses it is possible to relate the
concentration of drug in plasma to the level of target engagement. These data can
be fitted to an “Emax” model in order to estimate the drugs 𝐸𝐶!",
𝑂𝑐𝑐 = !!! !""!"#
!!!!!"!"! (6)
where 𝐸𝐶!" is the concentration of drug that leads to half maximal target
engagement, 𝑁 is the Hill coefficient describing cooperativity (𝑁 =1 for classical
antagonist drug binding) and 𝑂𝑐𝑐!"# is the maximal level of target engagement.
Figure 3: PET Target Engagement studies of four different molecular targets. Top
rows illustrate baseline scans and bottom rows illustrate the same subject imaged
GlyT1 – [11C]GSK931145 Mu Opioid – [11C]Carfentanil
Histamine H3 – [11C]GSK189254 SERT – [11C]DASB
after the administration of a drug candidate. Reduction in the overall signal and
heterogeneity confirms that the drug crosses the BBB and behinds to that molecular
target.
Target Engagement +
Classical target engagement studies have looked at a single time point post drug
dosing and vary only the administered dose of the drug. If the drug kinetics are
“Direct”, meaning that the brain free drug concentration can be assumed to be in
equilibrium with the free plasma concentration at all times, and that the drug-target
residence time is suitably finite, then the 𝐸𝐶!" can be readily estimated from such
experiments. If these conditions are not met and the drug kinetics are “Indirect” then
improved experimental designs are required which involve the measurement of
occupancy at different time points post-administration and varying doses [25-27].
Such experimental designs allow for an assessment of whether an Indirect
relationship is present and if it is, the modelling of the drug PK – target occupancy
can be achieved with a more sophisticated model that accounts for these kinetics.
! !""(!)! !
= 𝑘!" 𝐶! 𝑡 1 − 𝑂𝑐𝑐(𝑡) − 𝑘!"" 𝑂𝑐𝑐 𝑡 (7)
where 𝑘!" and 𝑘!"" are the target association and dissociation rate constants,
respectively, and 𝐸𝐶!" ( = 𝑘!""/𝑘!") which is equivalent to the drug plasma
concentration that achieves half maximal occupancy.
Figure 4. Examples of Direct and Indirect drug PK – target occupancy kinetics. (left)
Direct PK-RO kinetics are described by a classical Emax model independent of the
time at which target occupancy measures are taken, (right) Indirect PK-RO kinetics
exhibit hysteresis where later target occupancy measures have a higher occupancy
value in relation to plasma levels in contrast to measures at earlier time points such
that an Emax model no longer describes the relationship.
Whilst early studies used fixed designs with a pre-set range of doses, the latest
studies benefit from adaptive designs that use data from the study as it proceeds to
efficiently calculate the optimal doses and scan timings . This means that the
information acquired from scans is maximised and that study size and duration can
be minimized [25, 27]. Typically, initial doses and timings are obtained from
preclinical data and information on the drugs PK. Following the acquisition of a first
cohort of human subjects (usually 2 are sufficient) D-optimal design theory allows
doses and timings to be selected for the subsequent cohort. This process continues
throughout the study until the required precision has been obtained on the 𝐸𝐶!".
FTIH target engagement studies are all performed following a single dose (SD) of
drug. However, drugs are usually ultimately given to patients as a repeat dosing
Plasma Concentration10-1 100 101 102 103
Occ
upan
cy
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3 hr4 to 27 hrs32 to 61 hrs
Plasma Concentration10-3 10-2 10-1 100 101
Occ
upan
cy
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
up to 4 hr10 to 16 hrs23 to 33 hrs
Direct PK-RO relationship Indirect PK-RO relationship
(RD) either once or twice daily and therefore a drug development team is actually
interested in knowing what level of target engagement is reached at RD. If the drug
PK – occupancy relationship is Direct then the 𝐸𝐶!" measured from a SD PET study
will translate to an RD study and only RD PK is needed. However, if the relationship
is Indirect then one either needs to measure the occupancy from a RD PET study or
predict it from the SD PET study using an appropriate mathematical model [25, 26].
Figure 5. Predicting RD drug PK – RO relationships from SD data. Biomathematical
characterisation of the relationship between drug PK and brain target occupancy
from a SD PET study through modelling of discrete measures (white dots) of drug
PK and PET occupancy (top). This model may then be applied to RD drug PK data
to predict RD brain target occupancy (bottom).
The Indirect model described in equation 7 has been successfully applied to
Duloxetine to predict its RD occupancy at the serotonin transporter from SD PET
data [25]. A direct measurement of RD occupancy by conducting a RD PET study
carries the significant risk the up- or down-regulation of the molecular target is
induced by multiple dosing of the drug. Recent developments by Rabiner et al. (in
preparation) show that the parsimonious explanation for consistent differences in
RD Plasma Concentration RD Occupancy PK
– R
O M
odel
SD Occupancy SD Plasma Concentration
SD RD PK Model
𝐸𝐶!" estimates obtained from SD and RD occupancy studies for D2 receptor
antagonists is upregulation of the D2 receptor between the two PET measures.
Under such conditions, and assuming the administration of the drug does not change
the target affinity for either the drug or the radioligand, the 𝐸𝐶!" can be obtained
along with an estimate of the level of upregulation from,
𝑂𝑐𝑐 = !! !! ! !"!"
1 − !!"#!! !"!"
!! ! !"!" (8)
where U is the level of upregulation (U=1 corresponds to no change in target
number).
Figure 6. Upregulation of target concentration between baseline and post-drug
follow up scan leads to incorrect estimation of 𝐸𝐶!" if it is not properly accounted for.
Upregulation of the target, for example through repeat dose antagonist stimulation,
will lead to an underestimation of the true occupancy and hence underestimate the
potency of the drug candidate.
Plasma Concentration10-1 100 101 102 103
Occ
upan
cy
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Measured OccupancyTrue Occupancy
Pharmacological Activity
Estimation of pharmacological activity with PET aims to detect treatment induced
changes “downstream” of target engagement, though its utility in drug development
is not well established. [15O]H2O and 18F-FDG have been used in the past to
measure cerebral blood flow and brain glucose utilisation as indices of brain activity,
but have been supplanted by functional MRI for this purpose [28]. Irrespective of the
methodology used, the link between novel drug administration and changes in brain
activity has not been established with sufficient specificity to provide confidence for
go-no-go decision making. 18F-FDG PET imaging of the brain may find utility in the
development of treatments for conditions such as Alzheimer’s disease [29], but more
work remains to be done in demonstrating its utility for drug development.
Alternative approaches to imaging pharmacodynamics responses to drug
administration include the indexing of neuroinflammation (e.g. using a variety of
radioligands for the 18 kDa translocator protein – TSPO – to index microglial
activation) [30], evaluating the changes in misfolded protein deposition (using Ab and
tau radioligands) [10, 31] and assaying changes in synaptic neurotransmitter release
[32].
In the last decade, radioligands have been developed to quantify misfolded protein
concentrations in the brain following the seminal work of Klunk and Mathis in
Pittsburgh who introduced [11C]PiB for imaging amyloid in the brain [33]. Three
fluorinated analogues that are FDA approved for imaging Aβ and a number of tau
tracers are in development including [18F]AV1451, [18F]GE5351 and [11C]PBB3 [34-
36].
Aβ agents have already been employed in clinical trials to quantify the level of Aβ in
the brain pre- and post-treatment with novel drug candidates [10]. These imaging
tools are also being use to stratify entry of subjects into clinical trials where an Aβ+
status gives increased confidence that the subject is part of the target trial
population.
Neuroinflammation is a complex physiological process present in a variety of
neuropsychiatric diseases, including neurodegenerative disorders, viral infections
such as HIV and HTLV-1 and mood disorders. Molecular imaging of
neuroinflammation offers the prospect of assessing disease progression and the
effect of medication. The 18 kDa translocator protein (TSPO) has been studied
extensively as molecular marker of activated microglia [37, 38]. The development of
2nd generation PET ligands such as [11C]PBR28, [18F]DPA-713 and [18F]FEPPA
has provided improved tools to quantify TSPO availability, compared to the original
TSPO ligand [11C]PK11195 [39-42]. The high variability in binding of these ligands
has been addressed by the discovery of the rs6971 SNP in the TSPO gene, which
allows an efficient method to prescreen study populations to reduce experimental
variability [43]. [11C]PBR28 has been used recently to index the pharmacodynamic
effects of a novel myeloperoxidase inhibitor AZ3241 [44].
Assessment of synaptic neurotransmission processes has been enabled for the
dopamine system through the development of radioligands such as [11C]raclopride
and [11C]PHNO, that are sensitive to changes in endogenous dopamine levels with
radioligands [32, 45]. These tools have been used to show enhanced dopamine
release in schizophrenics in the striatum [46, 47], but not in extra-striatal areas [48].
Similar techniques have demonstrated altered opioid neurotransmission in
pathological gamblers [49, 50] and patients with alcohol dependence. Recent
developments offer the prospect of these techniques being extended to the
acetylcholine [51, 52], GABA [53] and serotonin [54, 55]. Assessment of changes in
evoked neurotransmitter release offers the means to assess the effect of drug
therapies on these neurotransmitter systems.
Pharmacological Activity +
Imaging tau is rapidly evolving with putative agents for this only being introduced in
the last few years and their characterization and validation is still under way.
However, there is already a strong interest in using these imaging tools in clinical
trials with an understanding that tau levels may be more closely linked with cognitive
function and the potential for more relevant readouts in trials evaluating tau targeted
therapies [34, 56].
Biomarker Development
Radiolabelling of a drug candidate for a biodistribution study requires the
radiochemistry expertise to attach a suitable positron emitting radionuclide to the
drug candidate and therefore the challenges are restricted to a problem in
radiochemistry. The development of a successful PET biomarker is a much more
challenging problem, which whilst requiring the equivalent radiochemistry expertise
also includes many other complexities surrounding the compound to radiolabel. This
is like a mini drug development programme in itself with the compound needing
appropriate characteristics such that it readily crosses the BBB, binds with high
enough affinity and selectivity to the biological target, low enough non-specific
binding so background signal does not dominate and suitable kinetics so that an
appropriate outcome measure can be estimated [57, 58]. Since the first development
of radioligands for GPCR targets in the late 80’s, initial screening of compounds has
focussed on identifying those with suitable lipophilicty (typically Log P 1-3) to ensure
BBB penetration and nM or better affinity to yield signal at the target. Radiolabelling
feasibility and pre-clinical evaluation follow prior to translation and evaluation in man.
To date, their exists around 30 protein targets for which successful PET radioligands
have been developed (see Table 1) [59].
Target System Description Radioligand
Dopamine D2/3 receptor antagonist
[11C]raclopride [18F]fallypride [11C]FLB457
D2/3 receptor agonist [11C]-(+)-PHNO [11C]NPA
D1 receptor antagonist [11C]NNC112 [11C]SCH23390
DAT antagonist [11C]PE2I [11C]CFT
Substrate for AADC [18F]DOPA [18F]FMT
VMAT2 antagonist [11C]DTBZ [18F]AV133
Serotonin SERT antagonist
[11C]DASB [11C]AFM
5-HT1A receptor antagonist [11C-carbonyl]WAY100635 [18F]FCWAY [11C]CUMI101
5-HT2A receptor antagonist [11C]MDL100907 [18F]Altanserin
5-HT2A receptor agonist [11C]CIMBI-36
5-HT1B receptor antagonist [11C]P943 [11C]AZ10419369
5-HT6 receptor antagonist [11C]GSK215083 Norepinepherine
NET antagonist [18F]MeNER [11C]MRB
Opioid m opioid receptor agonist [11C]carfentanil Non-selective opioid receptor antagonist [11C]diprenorphine
Acetylcholine α4β2 nAChR receptor antagonist 2-[18F]-FA-85380 M2 receptor agonist [18F]FP-TZTP
Glycine GlyT1 antagonist
[11C]GSK931145 [18F]MK-6577
GABA GABAA receptor antagonist [11C]flumazenil Glutamate allosteric mGluR5 receptor
antagonist [11C]ABP688 [18F]FPEB
Substance P NK1 receptor antagonist [18F]SPA-RQ Enzymes MAO B antagonist [11C]deprenyl
MAO A antagonist [11C]clorgyline PDE4 antagonist [11C]rolipram
PDE10 antagonist [11C]IMA107 [18F]MNI659
TSPO TSPO ligand
[11C]PK11195 [11C]PBR28 [18F]FEPPA
Cannabinoid CB1 receptor antagonist
[11C]OMAR [18F]MK9470
FAAH antagonist [11C]CURB β-Amyloid
β-sheet fibrils of β-Amyloid
[11C]PiB [11C]AZD4694 [18F]GE067 [18F]BAY949172 [18F]AV45
Tau PHF -tau
[18F]AV1451 [18F]THK5351 [11C]PBB3
Table 1. List of selected drug development targets and some of the PET radiotracers that allow their measurement.
Biomarker Development +
In recent years, the discovery and development of PET biomarkers has increased in
sophistication with the introduction of in silico biomathematical models that provide a
quantitative prediction of the compounds in vivo performance based on in vitro data
for initial screening of compound libraries [60, 61]. Such in silico methods can use
estimates of the molecular size and lipholicity to predict the delivery rate constant
(K1) in to brain tissue, affinity and non-specific binding assays (from equilibrium
dialysis assays) combined with target density information to enable simulation of
brain tissue kinetics in target and reference regions. From this it is possible to predict
in vivo signal to noise levels in the outcome measure of interest (usually the binding
potential – BPND) and thus rank candidates in terms of their likely performance in
vivo. Combining this with information on ease of radiolabelling based on the
compound structure allows for an early assessment about whether a compound
series contains candidates that would make PET imaging tractable and if so
prioritises the order in which compounds are selected for in vivo testing.
Analysis
Strategies for the acquisition and analysis of PET data range from the more complex
fully quantitative through to simpler semi-quantitative methods (e.g. SUVR). In all the
drug development applications discussed above (excepting SUVR for Aβ
stratification of subjects as Aβ -/ Aβ +), quantitative methods are necessary. For
example in Biodistribution studies, the use of sophisticated analysis methods allows
for the partitioning of the measured PET signal between tissue and blood, and for
therapeutic intervention studies it ensures that the changes in the outcome measure
relate to differences in target availability rather than blood flow.
Fully quantitative methods are comprised of blood data processing (if blood sampling
is performed), image processing, and kinetic modeling steps to derive regional or
parametric image estimates of the outcome measure of choice. Such analysis
workflows include image registration methods to correct for subject motion and align
anatomical information from co-acquired MR scans, along with tracer kinetic
modeling methods that model the kinetic behavior accounting for all the relevant
kinetics in the system and estimation of outcome measures that are directly related
to the target biology under investigation [59].
Analysis +
Quantitative imaging analysis has historically required many human steps with
several software packages and ‘home brew’ scripts needed to perform the full
analysis workflow from acquired data to end results. The software and workflow
employed, along with their myriad options are often detailed in individuals’ notebooks
and neither controlled nor entirely reproducible. This can lead to inefficiency,
inconsistency through human error, and end results that are difficult or impossible to
reproduce later.
Quality, consistency and reproducibility of PET analysis are entirely attainable, and
software tools now exist to facilitate this (MIAKAT; www.miakat.org). Such software
tools provide complete, audit trailed analysis of the data allowing for transparency on
what analyses have been applied and allowing replication of the analysis from the
primary data at any point. Such advances in analysis tools provide increased
confidence in the analysis of PET drug development studies.
Summary
In summary, PET molecular imaging provides a unique window into the brain for
CNS drug hunters (Figure 7). It allows direct assays of the free concentration of the
drug in brain tissue, the level of target engagement at particular administered doses
and the level of downstream pharmacological activity. These three pillars of drug
survival can all be assessed early in FTIH, in small studies (N~10) providing critical
decision making data for early discharge of risk. It should be noted that the same
techniques can be applied in early preclinical testing, where these three pillars can
also be assessed. Finally, PET molecular imaging is taking on an ever increasing
role in later phase CNS studies in neurodegenerative diseases, where Aβ scans are
now used routinely for the stratification of subjects into clinical trials and for providing
readouts on therapeutic interventions.
Figure 7. Summary of where PET imaging impacts across the Drug Development
pipeline.
Tissue&Exposure&
Target&Engagement&
Pharmacological&Ac6vity&
Pa6ent&Stra6fica6on&
Disclosure/Conflict of Interest
RG is a consultant for Abbvie, GlaxoSmithKline, and UCB.
ER is a consultant for Opiant Pharmaceuticals and GlaxoSmithKline.
References
1. Hargreaves RJ, Hoppin J, Sevigny J, et al: Optimizing Central Nervous System Drug Development Using Molecular Imaging: Clin Pharmacol Ther 2015; 98:47-‐60.
2. Morgan P, Van Der Graaf PH, Arrowsmith J, et al: Can the flow of medicines be improved? Fundamental pharmacokinetic and pharmacological principles toward improving Phase II survival: Drug Discov Today 2012; 17:419-‐24.
3. Gunn R, Rabiner I: Making drug development visible -‐ and viable: Drug Discovery Today 2014; 19:1-‐3.
4. Farde L, Wiesel FA, Halldin C, et al: Central D2-‐dopamine receptor occupancy in schizophrenic patients treated with antipsychotic drugs: Arch Gen Psychiatry 1988; 45:71-‐6.
5. Bench CJ, Lammertsma AA, Dolan RJ, et al: Dose dependent occupancy of central dopamine D2 receptors by the novel neuroleptic CP-‐88,059-‐01: a study using positron emission tomography and 11C-‐raclopride: Psychopharmacology (Berl) 1993; 112:308-‐14.
6. Hargreaves RJ, Rabiner EA: Translational PET imaging research: Neurobiol Dis 2014; 61:32-‐8. 7. Cunningham VJ, Parker CA, Rabiner EA, et al: PET studies in drug development:
Methodological considerations: Drug Discov Today Technol 2005; 2:311-‐5. 8. Matthews PM, Rabiner I, Gunn R: Non-‐invasive imaging in experimental medicine for drug
development: Curr Opin Pharmacol 2011; 11:501-‐7. 9. Papisov MI, Belov V, Fischman AJ, et al: Delivery of proteins to CNS as seen and measured by
positron emission tomography: Drug Deliv Transl Res 2012; 2:201-‐9. 10. Rinne JO, Brooks DJ, Rossor MN, et al: 11C-‐PiB PET assessment of change in fibrillar amyloid-‐
beta load in patients with Alzheimer's disease treated with bapineuzumab: a phase 2, double-‐blind, placebo-‐controlled, ascending-‐dose study: Lancet Neurol 2010; 9:363-‐72.
11. Gunn RN, Gunn SR, Cunningham VJ: Positron emission tomography compartmental models: J Cereb Blood Flow Metab 2001; 21:635-‐52.
12. Gunn RN, Summerfield SG, Salinas CA, et al: Combining PET biodistribution and equilibrium dialysis assays to assess the free brain concentration and BBB transport of CNS drugs: J Cereb Blood Flow Metab 2012; 32:874-‐83.
13. Summerfield SG, Lucas AJ, Porter RA, et al: Toward an improved prediction of human in vivo brain penetration: Xenobiotica 2008; 38:1518-‐35.
14. Jansen MH, Lagerweij T, Sewing AC, et al: Bevacizumab targeting diffuse intrinsic pontine glioma: results of 89Zr-‐bevacizumab PET imaging in brain tumor models: Mol Cancer Ther 2016.
15. Luo H, Hernandez R, Hong H, et al: Noninvasive brain cancer imaging with a bispecific antibody fragment, generated via click chemistry: Proc Natl Acad Sci U S A 2015; 112:12806-‐11.
16. Altai M, Perols A, Tsourma M, et al: Feasibility of Affibody-‐Based Bioorthogonal Chemistry-‐Mediated Radionuclide Pretargeting: J Nucl Med 2016; 57:431-‐6.
17. Ashworth S, Berges A, Rabiner EA, et al: Unexpectedly high affinity of a novel histamine H(3) receptor antagonist, GSK239512, in vivo in human brain, determined using PET: Br J Pharmacol 2014; 171:1241-‐9.
18. Brooks DJ, Doder M, Osman S, et al: Adenosine A(2A) receptor occupancy by istradefylline. An C-‐11 KW-‐6002 PET study in healthy subjects: Neurology 2005; 64:A235-‐A235.
19. Castner SA, Murthy NV, Ridler K, et al: Relationship Between glycine Transporter 1 Inhibition as Measured with Positron Emission Tomography and Changes in Cognitive Performances in Nonhuman Primates: Neuropsychopharmacology 2014.
20. Comley RA, Salinas CA, Slifstein M, et al: Monoamine transporter occupancy of a novel triple reuptake inhibitor in baboons and humans using positron emission tomography: J Pharmacol Exp Ther 2013; 346:311-‐7.
21. Rabiner E, Bhagwagar Z, Gunn R, et al: Attenuation of preferential occupancy of 5-‐HT1A autoreceptors by pindolol in depressed patients: effect of SSRIs or an endophenotype of the depressed state?: Neuroimage 2002; 16:S67-‐S67.
22. Rabiner EA, Gunn RN, Wilkins MR, et al: Evaluation of EMD 128 130 occupancy of the 5-‐HT1A and the D-‐2 receptor: a human PET study with C-‐11 WAY-‐100635 and C-‐11 raclopride: Journal of Psychopharmacology 2002; 16:195-‐199.
23. Ridler K, Gunn RN, Searle GE, et al: Characterising the plasma-‐target occupancy relationship of the neurokinin antagonist GSK1144814 with PET: J Psychopharmacol 2014; 28:244-‐53.
24. Searle GE, Beaver JD, Tziortzi A, et al: Mathematical modelling of [(1)(1)C]-‐(+)-‐PHNO human competition studies: Neuroimage 2013; 68:119-‐32.
25. Abanades S, van der Aart J, Barletta JA, et al: Prediction of repeat-‐dose occupancy from single-‐dose data: characterisation of the relationship between plasma pharmacokinetics and brain target occupancy: J Cereb Blood Flow Metab 2011; 31:944-‐52.
26. Salinas C, Weinzimmer D, Searle G, et al: Kinetic analysis of drug-‐target interactions with PET for characterization of pharmacological hysteresis: J Cereb Blood Flow Metab 2013; 33:700-‐7.
27. Zamuner S, Di Iorio VL, Nyberg J, et al: Adaptive-‐optimal design in PET occupancy studies: Clin Pharmacol Ther 2010; 87:563-‐71.
28. Borsook D, Becerra L, Hargreaves R: A role for fMRI in optimizing CNS drug development: Nat Rev Drug Discov 2006; 5:411-‐24.
29. Cummings JL, Banks SJ, Gary RK, et al: Alzheimer's disease drug development: translational neuroscience strategies: CNS Spectr 2013; 18:128-‐38.
30. Garvey LJ, Pavese N, Politis M, et al: Increased microglia activation in neurologically asymptomatic HIV-‐infected patients receiving effective ART: AIDS 2014; 28:67-‐72.
31. Wang L, Benzinger TL, Su Y, et al: Evaluation of Tau Imaging in Staging Alzheimer Disease and Revealing Interactions Between beta-‐Amyloid and Tauopathy: JAMA Neurol 2016.
32. Laruelle M: Imaging synaptic neurotransmission with in vivo binding competition techniques: a critical review: J Cereb Blood Flow Metab 2000; 20:423-‐51.
33. Klunk WE, Engler H, Nordberg A, et al: Imaging brain amyloid in Alzheimer's disease with Pittsburgh Compound-‐B: Ann Neurol 2004; 55:306-‐19.
34. Ossenkoppele R, Schonhaut DR, Scholl M, et al: Tau PET patterns mirror clinical and neuroanatomical variability in Alzheimer's disease: Brain 2016; 139:1551-‐67.
35. Lockhart SN, Baker SL, Okamura N, et al: Dynamic PET Measures of Tau Accumulation in Cognitively Normal Older Adults and Alzheimer's Disease Patients Measured Using [18F] THK-‐5351: PLoS One 2016; 11:e0158460.
36. Kimura Y, Ichise M, Ito H, et al: PET Quantification of Tau Pathology in Human Brain with 11C-‐PBB3: J Nucl Med 2015; 56:1359-‐65.
37. Vivash L, O'Brien TJ: Imaging Microglial Activation with TSPO PET: Lighting Up Neurologic Diseases?: J Nucl Med 2016; 57:165-‐8.
38. Airas L, Rissanen E, Rinne JO: Imaging neuroinflammation in multiple sclerosis using TSPO-‐PET: Clin Transl Imaging 2015; 3:461-‐473.
39. Kreisl WC, Fujita M, Fujimura Y, et al: Comparison of [(11)C]-‐(R)-‐PK 11195 and [(11)C]PBR28, two radioligands for translocator protein (18 kDa) in human and monkey: Implications for positron emission tomographic imaging of this inflammation biomarker: Neuroimage 2010; 49:2924-‐32.
40. Kreisl WC, Lyoo CH, Liow JS, et al: (11)C-‐PBR28 binding to translocator protein increases with progression of Alzheimer's disease: Neurobiol Aging 2016; 44:53-‐61.
41. Rusjan PM, Wilson AA, Bloomfield PM, et al: Quantitation of translocator protein binding in human brain with the novel radioligand [18F]-‐FEPPA and positron emission tomography: J Cereb Blood Flow Metab 2011; 31:1807-‐16.
42. Yokokura M, Terada T, Bunai T, et al: Depiction of microglial activation in aging and dementia: Positron emission tomography with [11C]DPA713 versus [11C](R)PK11195: J Cereb Blood Flow Metab 2016.
43. Owen DR, Yeo AJ, Gunn RN, et al: An 18-‐kDa translocator protein (TSPO) polymorphism explains differences in binding affinity of the PET radioligand PBR28: J Cereb Blood Flow Metab 2012; 32:1-‐5.
44. Jucaite A, Svenningsson P, Rinne JO, et al: Effect of the myeloperoxidase inhibitor AZD3241 on microglia: a PET study in Parkinson's disease: Brain 2015; 138:2687-‐700.
45. Shotbolt P, Tziortzi AC, Searle GE, et al: Within-‐subject comparison of [(11)C]-‐(+)-‐PHNO and [(11)C]raclopride sensitivity to acute amphetamine challenge in healthy humans: J Cereb Blood Flow Metab 2012; 32:127-‐36.
46. Abi-‐Dargham A, Gil R, Krystal J, et al: Increased striatal dopamine transmission in schizophrenia: confirmation in a second cohort: Am J Psychiatry 1998; 155:761-‐7.
47. Laruelle M, Abi-‐Dargham A, Gil R, et al: Increased dopamine transmission in schizophrenia: relationship to illness phases: Biol Psychiatry 1999; 46:56-‐72.
48. Slifstein M, van de Giessen E, Van Snellenberg J, et al: Deficits in prefrontal cortical and extrastriatal dopamine release in schizophrenia: a positron emission tomographic functional magnetic resonance imaging study: JAMA Psychiatry 2015; 72:316-‐24.
49. Colasanti A, Guo Q, Giannetti P, et al: Hippocampal Neuroinflammation, Functional Connectivity, and Depressive Symptoms in Multiple Sclerosis: Biol Psychiatry 2016; 80:62-‐72.
50. Mick I, Myers J, Ramos AC, et al: Blunted Endogenous Opioid Release Following an Oral Amphetamine Challenge in Pathological Gamblers: Neuropsychopharmacology 2016; 41:1742-‐50.
51. Esterlis I, Hannestad JO, Bois F, et al: Imaging changes in synaptic acetylcholine availability in living human subjects: J Nucl Med 2013; 54:78-‐82.
52. Hillmer AT, Esterlis I, Gallezot JD, et al: Imaging of cerebral alpha4beta2* nicotinic acetylcholine receptors with (-‐)-‐[18F]Flubatine PET: Implementation of bolus plus constant infusion and sensitivity to acetylcholine in human brain: Neuroimage 2016; 141:71-‐80.
53. Frankle WG, Cho RY, Narendran R, et al: Tiagabine increases [11C]flumazenil binding in cortical brain regions in healthy control subjects: Neuropsychopharmacology 2009; 34:624-‐33.
54. Finnema SJ, Varrone A, Hwang TJ, et al: Fenfluramine-‐induced serotonin release decreases [11C]AZ10419369 binding to 5-‐HT1B-‐receptors in the primate brain: Synapse 2010; 64:573-‐7.
55. Ridler K, Plisson C, Rabiner EA, et al: Characterization of in vivo pharmacological properties and sensitivity to endogenous serotonin of [11C] P943: a positron emission tomography study in Papio anubis: Synapse 2011; 65:1119-‐27.
56. Brier MR, Gordon B, Friedrichsen K, et al: Tau and Abeta imaging, CSF measures, and cognition in Alzheimer's disease: Sci Transl Med 2016; 8:338ra66.
57. Honer M, Gobbi L, Martarello L, et al: Radioligand development for molecular imaging of the central nervous system with positron emission tomography: Drug Discov Today 2014; 19:1936-‐44.
58. Pike VW: Considerations in the Development of Reversibly Binding PET Radioligands for Brain Imaging: Curr Med Chem 2016; 23:1818-‐69.
59. Gunn RN, Slifstein M, Searle GE, et al: Quantitative imaging of protein targets in the human brain with PET: Phys Med Biol 2015; 60:R363-‐411.
60. Guo Q, Brady M, Gunn RN: A biomathematical modeling approach to central nervous system radioligand discovery and development: J Nucl Med 2009; 50:1715-‐23.
61. Zhang L, Villalobos A, Beck EM, et al: Design and selection parameters to accelerate the discovery of novel central nervous system positron emission tomography (PET) ligands and
their application in the development of a novel phosphodiesterase 2A PET ligand: J Med Chem 2013; 56:4568-‐79.