Modeling and simulation (M

1
Model Based Development of a Direct Factor Xa
Inhibitor S. Rohatagi1, T. Ozeki-Ishizuka2, Y.
Nitsu2, F. Ezzet3, H. Kastrissios3, T.J.
Carrothers3, S.J. Haworth1 1 Daiichi Sankyo
Pharma Development, Edison, NJ 2 Daiichi
Sankyo, Tokyo 3 - Pharsight Corporation,
Mountain View, CA
ABSTRACT
MODEL SCHEMATIC
OBJECTIVES
PROJECTED COMPARISON
Modeling and simulation (MS) was employed to
recommend doses for human Phase I studies of a
direct Factor Xa (FXa) inhibitor, CS-3030.
Predicted human pharmacokinetics (PK), biomarker
responses (PD), and clinical outcomes were
obtained using appropriate projection methods and
PK/PD data from cynomolgus monkey, together with
literature data. Models were developed for
anti-FXa activity and fold-increase in
prothrombin time (PT) compared to baseline using
the following criteria to determine the target
dose range (1) anti-FXa activity within 0.5-0.8
IU/mL range (based on enoxaparin) and (2) 2- to
3-fold increase in PT (based on warfarin). PK/PD
for a range of CS-3030 doses (10 to 320 mg),
regimens (single dose, once daily (QD) and twice
daily (BID)) and bioavailability fractions (4.5
to 50) were simulated. The ranges of doses and
bioavailability fractions were intended to
compensate for any misspecification due to
projection method or underlying assumptions.
Influences of patient demographics and laboratory
values were investigated on response to CS-3030.
No one dose met the dual criteria of anti-FXa
activity and PT response. Rather, target levels
were achieved only partially over the dosing
interval. If a single criterion was used, e.g.
anti-FXa activity only, then a dose of 40 mg
provided 50 time within the targeted range.
Renal impairment was expected to influence drug
exposure, and the effect was smaller for PT
response than anti-FXa activity. Appropriate dose
adjustment is thus possible for different
populations. Human projections from animal FXa
activity suggest doses up to 40 mg/day CS-3030
may provide similar efficacy (prevention of deep
vein thrombosis) and safety (risk of bleeding)
profiles to that of enoxaparin doses up to 100
mg/day following hip and knees surgeries. In
conclusion, MS led to identification of key
elements to be studied earlier than usual, i.e.,
bioavailability and the effect of renal
clearance, and the FIM study could be designed
accordingly. This illustrated the application of
MS to guide drug development and inform the
design of clinical trials.
Comparison across drugs shows favorable projected
response profile for CS-3030, comparable to
observed responses for comparator drugs.

• Modeling and simulation (MS) were employed to
  make dosage recommendations for human Phase 1
  studies of CS-3030.
• Specific objectives were to
• Predict human PK-PD of CS-3030 based on animal to
  human projections,
• Characterize sources of variability or safety
  concerns, and
• Simulate potential clinical outcomes as compared
  to other anticoagulants.

BACKGROUND
MOTIVATION
CS-3030 is an oral, direct Factor Xa (FXa)
inhibitor in development for the management of
thromboembolic diseases. In animal studies,
CS-3030 is cleared largely by the kidney. It is
not metabolized by CYP 450 isozymes and therefore
is expected to have low potential for drug-drug
interactions. An International Normalized Ratio
(INR) of 2-3 fold is generally considered a safe
and effective anticoagulant range, thereby
serving as a practical guide to dose selection
for clinical use. Prothrombin time (PT)
prolongation is known to be mediated by FXa
inhibition therefore, determination of PT and
anti-FXa activity during preclinical development
provides a basis for driving the drug development
process towards selection of doses associated
with target anti-FXa activity and PT/INR range.
SUMMARY OF RESULTS
None of the doses met the dual criteria of
anti-FXa activity and PT response. Rather, target
levels were achieved only partially over the
dosing interval. Proportions of the anti-FXa and
PT profiles within the targeted range were
consistently larger for BID regimens as compared
to QD regimens. If a single criterion was used,
e.g. anti-FXa activity only, then a dose of 40 mg
provided 50 time within the targetted range.
Renal impairment was expected to influence drug
exposure, and therefore PD. The effect was
smaller for PT response than anti-FXa
activity. For a subject with severe renal
impairment, average anti-FXa activity was
approximately double that a healthy subject. This
may suggest that appropriate dose adjustment may
be warranted if target anti-FXa activity were to
be maintained close to target values. Human
projections from animal FXa activity suggest
doses up to 40 mg/day CS-3030 may provide similar
efficacy (prevention of deep vein thrombosis) and
safety (risk of bleeding) profiles to that of
enoxaparin following hip and knees
surgeries. However, doses of 10 80 mg show
lower bioavailability and large intersubject
variability.
RESULTS PK/PD MODELING
Simulated PK Profiles in Humans Allometric
Scaling from Monkey Data
Simulated PK and Biomarker Profiles Special
Populations
METHODS

• Predicted human pharmacokinetics (PK), biomarker
  responses (PD), and clinical outcomes were
  obtained using appropriate projection methods and
  PK/PD data from cynomolgus monkey, relative
  potency data and literature data.
• Allometric scaling was used to predict human
  pharmacokinetics.
• Models were developed for anti-FXa activity and
  fold-increase (i.e., multiples of the baseline
  value) in PT using the following criteria to
  determine the target dose range
• anti-FXa activity within 0.5-0.8 IU/mL range
  (based on enoxaparin)
• 2- to 3-fold increase in PT (based on warfarin)
• It was assumed that PK-PD relationships observed
  in cynomolgus monkeys apply to humans.

• What do we do now?
• We model the drug as part of analysis.
• We are reactive
• What should we do?
• Think prospectively
• Provide the context for evaluating New Chemical
  Entities (NCEs)
• Inform key multi-faceted development decisions
• Product profiles vs. key competitors
• Treatment opportunities (mono vs. combo, doses)
• Special populations and covariates
• Communicate uncertainty in these attributes to
  decision makers
• Support informed Go/No-Go decision-making
• Ideally, critical development decisions should
  leverage relevant public and proprietary data
• Make model the basis of developing drugs
• When do we start?
• As early as possible
• What types of models
• PK/PD
• Disease models

CONCLUSIONS AND DISCUSSION

• Integration of animal data and public literature
  allowed human PK-PD to be projected under certain
  plausible assumptions and scenarios. Human
  projections for CS-3030 identified dosing
  regimens which provided similar efficacy and
  safety profiles to that of comparators
• MS was used to optimize the Phase 1 program to
  reduce uncertainty and test assumptions relating
  to bioavailability and variability, and further
  to provide a basis to
• Estimate the likely quality of Phase 2
  dose-response in Phase 1 planning.
• Quantify the effect of covariates, the magnitude
  and sources of uncertainty, and key assumptions.
• This example illustrates the application of MS
  to guide drug development and inform the design
  of clinical trials.

Simulated Biomarker Profiles in
Humans Anti-Factor Xa Activity and PT
Relationship Between Dose and Event Probability
PK/PD for a range of CS-3030 doses (10 to 320
mg), regimens (single dose, once daily (QD) and
twice daily (BID)), and bioavailability fractions
(4.5 to 50) were simulated. Ranges of doses and
bioavailability fractions were intended to
compensate for any misspecification due to
projection method or underlying
assumptions. Influences of patient demographics
and laboratory values were investigated on
response to CS-3030. Comparison of clinical
events was made using publicly available
literature for three comparators warfarin,
enoxaparin sodium (Lovenox) and fondaparinux
sodium (Arixtra).