Transcript PPT - ACoP

BACKGROUND
METHODS
Drug Development Decision Rules
Product Profile Maps
As a key component of model based drug development
(MBDD), quantitative decision criteria (QDC) provides
the important link between evidence synthesis (metaanalysis and/or PK-PD-disease progression modelling)
and the assessment of future trial performance (1).
Application within the MBDD framework across learning
and confirming phases of development has recently
been more thoroughly examined (2-3).
Model based meta-analysis (MBMA) of literature
data (4,5) and additional relevant data from Pfizer
internal studies were used to develop product profile
maps and enable the determination of target values
used in the QDC.
PH-797804 & Design
PH-797804 is a potent and selective orally active
inhibitor of p38 mitogen activated protein (MAP) kinase
which regulates the production and downstream
signaling of pro-inflammatory cytokines and has a role
in regulating neuronal plasticity and pain sensitisation.
In evaluating its potential in the treatment of chronic
NcP, a POC study has been designed to evaluate its
efficacy (in presence and absence of naproxen) in
relieving pain in patients with osteoarthritis (OA) of the
knee. Primary endpoint WOMAC pain subscale 0-10
numeric rating scale (NRS).
Efficacy
Priors for PH-797804
Operating Characteristics (OC)
Both conditional (based on assumed ∆) and
unconditional (integrated across P(∆|(prior data &
translational uncertainty)) operating characteristics
were evaluated across a range of design
assumptions (n, SD and Prior). While the two
decision rules could be optimized to provide a go
decision based a similar observed value, the
operating characteristics were shown to be
dependent on the design (sample size) and the
assumed
P(∆|(prior data & translational uncertainty).
[1] Lalonde RL .etal (2007) Model-based drug development Clinical
Pharmacol Ther 61:275-291
[2] Smith MK, etal. (2011) Decision-Making in drug development:
Application of a model based framework for assessing trial performance P
61- 83 Clinical Trial Simulations, AAPS Advances in the pharmaceutical
Sciences series 1. Kimko H.H.C and Peck.C.C (eds), Springer.
[3] Chuang stein etal (2011) A quantitative Approach for making go/no-go
decisions in drug development Drug Information Journal (45)187-202
[4] Rappaport BA, etal . ACTION on the prevention of chronic pain after
surgery. Anesthesiology 2010;112:509–10.
[5] Dworkin RH, etal Evidence-based clinical trial design for chronic pain
pharmacotherapy:A blueprint for ACTION PAIN (in press Available online
9th December 2010)
[6]Boucher M, A Bayesian Meta-Analysis of Longitudinal Data in Placebo
Controlled Studies with Naproxen, PAGE 2008. http://www.pagemeeting.org/pdf_assets/8269-Bayesian%20Meta%20Analysis%20Final.pdf
[7] Ezzet F Modeling Adverse Event Rates of Opioids
for the Treatment of Osteoarthritis Pain using Literature Data.PAGE 2010
http://www.page-meeting.org/pdf_assets/1411OA%20Pain%20PAGE%20Poster%202010.pdf
**The authors would like to thank Martin Boucher and Tracy Higgins
Pfizer Ltd for the meta-analyses which underpin the work presented
-1.5
-0.6
Optimistic Prior
PH-797804
-2.0
-1.8
Realistic Prior
PH-797804
density
0.005
0.010
0.015
0.030
0.025
0.020
0.0
-4
-2
0
2
4
6
-6
-4
-2
TrueTrue
Difference
Value
0
2
4
6
TrueTrue
Difference
Value
Conditional OCs
a)Pr(∆> SOC)  95%
b) Pr(∆> SOC*1.75)  25%
1.75*SOC
SOC
SOC
Go
Amber
(notable
analgesia)
Stop
True Difference
True Difference
Minimum Observed Mean value for Go (SD=1.5)
Unconditional OCs
a) (Pr(∆> SOC)  95%)
b) (Pr(∆> SOC*1.75)  25%)
Realistic Prior
Optimistic Prior
PPV= Positive predictive value NPV= Negative predictive value
Evaluation of Trial Sensitivity for Go
Decision
A sensitivity of the proposed design and decision rules
(b) to detect an increase in treatment effect over SOC
across baseline values, placebo/SOC efficacy and
background rate of non-response was undertaken. A
non-response/dropout rate of 30% was assumed
based on past experience. It was determined that the
design was able to detect a mean change required for
a Go decision (Pr(∆> SOC*1.75)  25%) provided the
inclusion criteria target a population with a mean
baseline NRS score of > 6 (moderate to severe pain
population).
DISCUSSION
-2.5
-1.2
RA to OA Translational
uncertainty
0.015
-6
Evidence of notable analgesic efficacy was defined
as: Pr(∆> placebo)  90% & Pr(∆> SOC)  20%
Two competing decision rules governing the Go
Decision, progression of PH-7978040 to a dose
ranging study, were evaluated:
a) High confidence of greater efficacy over SOC
(Pr(∆> SOC)  95%)
b) Lower confidence of achieving the TV which is
superior to SOC (Pr(∆> SOC*1.75)  25%).
The within Subject SD was estimated to 1.5 from
previous trials. Sample sizes of 16 (short dash line),
33 (solid-line), 66(long dash line) per group are
compared.
TV 1.58
Patients on NSAIDs
Change in WOMAC Pain Score
(0-10 scale) **
0.0
QDC
Oxycodone ER
-1.0
0.010
A previous RA POC study had been conducted for
PH-797804 and a break down of treatment effect by
background NSAID usage was available. Prior data
on NSAID and COX2 indicated that pain efficacy in
RA and OA patients were approximately similar. A
combination of this information with the meta-analyses
for SOC were used to predict the expected
treatment effect for PH-797804 in OA patients
with/without NSAIDs. In addition, an empirical
estimate for the translational uncertainty associated
with a new MOA was factored in to develop a realistic
design prior.
Nevertheless, selection was largely influenced by the
philosophy of identifying a higher hurdle and allowing
progression to phase 2b provided there is reasonable
evidence to indicate that this could be achieved in
later development.
• To design a POC study for PH-797804 in patients
with OA of the knee with a high probability of making
the correct decision at the end of the trial.
•To highlight some of the considerations in the context
of developing robust quantitative decision criteria in the
assessment of novel therapies for the treatment of
chronic nociceptive pain (NcP).
-0.5
0.0
0.0
While the choice of rule b) over rule a) was supported
by the operating characteristics:
•Slightly better OC with larger sample sizes( n=66)
•Control of type 1 error across sample size
(Unconditional OC’s)
•Minimize probability of Amber outcome
OBJECTIVES
Patients not on other Tx
Change in WOMAC Pain Score
(0-10 scale) *
Naproxen
* 1.75
SOC= -0.9
Density
density
The challenge of low yield and generic competition is a
harsh reality of modern drug development. In this
regard, despite considerable investments, the hurdles
facing novel analgesic drugs are particularly high. For
the treatment of chronic nociceptive pain (NcP) there
remains limited therapeutic options, (essentially
acetaminophen, NSAID or opioids) and patients
continue to be inadequately treated. Differentiation from
these agents is a clear goal for early analgesic drug
development.
Efficient early stage screening is required in order to
prevent late stage clinical or commercial failure.
Quantitatively robust product concepts are needed to
provide the foundation for effective decision-making.
Criteria need to efficiently screen for analgesics which
have sufficient signal to warrant further investment;
where their potential can be enhanced upon identifying
the optimal dose regimen in a phenotypically targeted
patient population.
Target Range
Compound and Indication: PH797804 for Chronic musculoskeletal pain
Patient population: OA
All efficacy reported as PBO-corrected. Assume baseline pain score ~7 on 0-10 NRS
scale & placebo response of 1-2 pts
Frequency
Analgesic Drug Development Challenge
RESULTS
0.005

Developing quantitative decision rules for novel analgesic drugs:
A case study illustrating current experiences and learnings
Authors: Scott Marshall(1), Simon Kirby(1), Anna Crossland(1), Paul Sanders(1),
Joe Picard(2), and Bernadette Hughes(1).
Institutions: (1) Pfizer Ltd, Sandwich, Kent, UK; (2) Pfizer Ltd, New London, CT, USA
* EOT= End of treatment mean NRS score ** Based on analysis of previous parallel group trials
In response to the current decline in the approval of new analgesic medicines the FDA has recently launched the
Analgesic Clinical Trial Innovations, Opportunities, and Networks (ACTION) initiative with the aim ‘‘to streamline
the discovery and development process for new analgesic drug products for the benefit of the public health’’ (4).
Study design factors (e.g. increasing placebo response, loss of assay sensitivity) malign the drug development
process of novel analgesics. The concept of an evidence-based approach to the design of analgesic clinical trials
with the aim of first understanding the relationships between trial characteristics and outcome has been proposed
(5). This case study illustrates the role that quantitative decision criteria can have in enabling such scrutiny in
order to prevent future trial design failure. Thorough assessment of analgesic trial characteristics is similarly
required in order to ensure product concepts can be adequately tested. This work also highlights the clear role
that unconditional operating characteristics have in helping to assess proposed quantitative decision criteria.