Download Exposure Response

The time course of tumour size
changes with gemcitabine.
What can we learn about
pharmacology?
Nick Holford
Dept Pharmacology & Clinical Pharmacology
University of Auckland
The Workers
Lai San Tham3
Boon-Cher Goh1
Wei Peng Yong1
Ross A Soo1
Ling-Zhi Wang1
Soo-Chin Lee1
How-Sung Lee2
1Department
of Hematology-Oncology, National University Hospital, Singapore
2Department of Pharmacology, National University of Singapore
3Lilly-NUS Centre for Clinical Pharmacology, Singapore
Outline
• What does Pharmacology mean?
• Does Cancer need PKPD?
• Learning and Confirming
– A reminder
• Tumor response study
– A pharmacodynamic model for the time course of tumor
shrinkage in patients with ‘big cell’ lung cancer
– What can we learn about pharmacology?
– How can we use tumour size to predict survival?
Pharmacokinetics
Pharmacodynamics
Treatment
Rx =Recipe
or
Jupiter
Symbol
What does this mean?
Confirming or Learning?
Confirming
• Making sure
• Outcome Expected
• Analysis Assumptions Minimized
E.g. Randomized Treatment Assignment
• Questions for Drug Approval
– E.g.
• Is the drug effective?
• Can it be used safely in renal failure?
Learning
• Exploration
• Outcome Unexpected
• Assumption rich analysis
– E.g. PKPD model
• Questions for Drug Science
– E.g.
• How big an effect does the drug have?
• What is the clearance in renal failure?
Sheiner LB. Learning versus confirming in clinical drug development. Clinical Pharmacology & Therapeutics 1997;61(3):275-91
"Why We're Losing the War on Cancer"
Accelerating Anticancer Agent Development and
Validation Workshop June 20-22, 2007
Keynote Address: "Learning Too Little, Too Late:
Why We Need a New Paradigm for the Cancer
Clinical Trial"
Clifton Leaf
Former Executive Editor
Fortune
Resisting RECIST
Throwing away data?
Collaborative Effort
Singapore Study
•
Randomized, phase II trial
•
Carboplatin at AUC of 5mg/mL*min
given over one hour on day 1 of each
cycle prior to the gemcitabine
infusion
•
Study arms
–
–
•
gemcitabine 750 mg/m2 over 75
minutes (Arm A) on days 1 and 8
every 3 weeks x 6 cycles
gemcitabine 1000 mg/m2 over 30
minutes (Arm B) on days 1 and 8
every 3 weeks x 6 cycles
No differences in outcome (survival
or toxicity)
Soo RA, Wang LZ, Tham LS, Yong WP, Boyer M, Lim HL, et al. A multicentre randomised phase II study of carboplatin in
combination with gemcitabine at standard rate or fixed dose rate infusion in patients with advanced stage non-small-cell
lung cancer. Ann Oncol. 2006;17(7):1128-33.
Singapore Study of Tumour Size
• 56 treatment naïve patients with advanced ‘big
cell’ lung cancer treated with carboplatin and
gemcitabine
• 261 measurements of tumour size
– largest dimension of the primary tumour measured
from CT images using electronic calipers
– Used only for RECIST category 
• Measurements at protocol baseline, cycles 2, 4
and 6, and bimonthly
– Actual mean follow up only 3.5 months
Gemcitabine Pharmacology
• Gemcitabine (dFdC)
– Inactive pro-drug
• dFdCTP (gemcitabine triphosphate)
– Intracellular, active, tri-phosphate metabolite
• dFdU
– Major extracellular, inactive metabolite
Exposure Response
• Which Exposure Measure?
– Dose
• Cannot distinguish PK from PD causes of variation
– AUC
• Can be used to identify causes of between patient
variability through PK model linking Dose to AUC
– C(t)
• Can be used to predict schedule dependence
• But long computation times preclude practical
exploration of C(t) and response
8
36
6
27
Concentration Spikes
With Each Cycle
4
18
2
9
Slow Tumour Response
0
0
0
10
20
30
40
Week
Tumour Size
Drug
50
60
Drug
Tumour Size
Why Dose by Dose Concentration Time
Course Won’t Work
How to Describe Delayed Drug Response?
• Exposure has delayed effect
–
–
–
–
Time course of drug concentration is complete within a few hours
Binding of drug to DNA probably rapid
Effect of DNA damage on cell proliferation probably slow
Time course of tumour response takes weeks
• KPD Effect Compartment Model for Exposure
– What “apparent half-life” of drug would explain the effect time
course?
– Can be based on Dose or AUC
– C(t) not required
dExposure
  ln( 2) / T 1 / 2, effect  Exposure
dt
Tumour Size Model
Drug Effect on Tumour
Formation Rate
dSize 
1

  RateIn  Effect 
 Size   Size
dt
Tturnover


Tumour Turnover Kinetics
Simple Feedback
Semi-mechanistic
Natural history of tumour growth has rapid growth with asymptote
Feedback inhibits growth
Drug effect expresses pathophysiological mechanism
“KPD” Drug Effect and Tumour Turnover
• Effect assumed to slow rate of proliferation of new
tumour cells
Dose
Effect  1 
Effect
Compartment
dCe
 k1  ( Dose  Ce)
dt
KPD Model
ln( 2)
k1 
T1/ 2,effect
E max  Ce
Dose50  Ce
Drug Effect Model
Tumour Turnover
Compartment
dSize
 ( RateIn  Effect  k 2  Size )  Size
dt
Tumour Model
T1/2,effect describes delay in drug effect
Tturnover describes delay in tumour esponse
k2 
1
Tturnover
“KPD” plus Turnover
8
36
6
27
4
18
Tumour Size
2
9
0
0
0
10
20
30
40
50
Week
Tumour Size
Ae
Dose
60
Dose and Ae
Tumour Size
Effective Amount of Drug
Variability in
Gemcitabine Dose-Response
Why Stop Treatment?
Which Exposure Metric?
• No better fit with dFdCTP (or dfdU) compared to
gemcitibine AUC
• No better fit with individual AUC compared with
dose of gemcitabine
• Dose is the simplest exposure metric
Tumour Size Turnover and PD Parameters
BSV
%
95% CI
6.7
54.6
(5.7, 7.8)
Tturnover, Tumor turnover (cm x week)
10.7
24.7
(2.0, 17.7)
Dose50, Gemcitabine dose at 50%
tumor size shrinkage (mg)
3156
136
(536, 16417)
T1/2,effect, KPD Effect half-life (week)
2.5
29
(0.61, 12.0)
12%
-
(9, 16)
Parameter
Size0, tumor size at baseline (cm)
Final
Estimate
Residual Error
Proportional error
BSV=Between Subject Variability (apparent coefficient of variation)
95%CI=Empirical confidence interval from 1000 bootstraps
What Did We Learn About
Pharmacology?
• No evidence that differences in exposure time course
[C(t)] can influence tumour response
• No evidence that intracellular metabolite is better then
dose as a predictor of tumour response
– Anti-tumour mechanism may be different from haematological
toxicities
• Unable to learn about influence of dose and duration of
infusion
– Uninformative design
Didn’t learn very much!
What Could We Learn About
Pharmacology?
• Can quantitate individual sensitivity (ED50) and
time course (effect and tumour ‘half-lives’)
• Complements toxicity based models e.g. Friberg
myelosuppression model (optimal dosing?)
• Link to survival probability (Claret et al 2009, Wang
et al 2009)
FDA Model Linking Tumour
Size with Survival
Wang Y, Sung C, Dartois C, Ramchandani R, Booth BP, Rock E, Gobburu J. Elucidation of relationship
between tumor size and survival in non-small-cell lung cancer patients can aid early decision making in clinical
drug development. Clin Pharmacol Ther. 2009;86(2):167-74.
Wang (FDA) Tumour Size Model
“A model with mixed exponential-decay (shrinkage) and linear-growth
(progression) components described the time course of tumor change
where TSi(t) is the tumor size at time t for the ith individual, BASEi is
the baseline tumor size, SRi is the exponential tumor shrinkage rate
constant, and PRi is the linear tumor progression rate.”
Empirical model
No dose or exposure information used
Claret Tumour Size Model
Log growth for tumor (no asymptote unlike Gompertz)
D(t)*exp(-λ*t) is identical to the effect compartment model
R(t) ‘resistance’ function cannot be distinguished from drug elimination
Tumour Size and Survival
Wang/Claret
Tumour Size
At 8 weeks
FDA Advisory Committee for Pharmaceutical Sciences and Clinical Pharmacology Meeting, March 18–19,
2008. http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4351b1-01-FDA.pdf
Tumour Size Predictions
FDA Advisory Committee for Pharmaceutical Sciences and Clinical Pharmacology Meeting, March 18–19,
2008. http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4351b1-01-FDA.pdf
Tumour Size Improves
Prediction of Survival
FDA Advisory Committee for Pharmaceutical Sciences and Clinical Pharmacology Meeting, March 18–19,
2008. http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4351b1-01-FDA.pdf
Which Patient Will Survive Longer?
Wang et al. and Claret et al. use tumour size at a fixed time to predict survival.
These patients have the same 8 week tumour size but different response time course
Survival models should include time course of tumour size!
Way to go!
Backup
Commentary
Clinical Pharmacology and Therapeutics
1. “Drug-independent models that link biomarker response
to clinical end points are critical to support early (end of
phase II) clinical decisions.
2. In oncology, change in tumor size (a biomarker of drug
effect evaluated in phase II) is linked to survival (a
phase II end point) in some solid tumors.
3. Change in tumor size can be used as a primary end
point in the design and evaluation of phase II studies
and in supporting go/no-go decisions and phase II
study design.”
Bruno R, Claret L. On the use of change in tumor size to predict survival in clinical oncology studies: toward a
new paradigm to design and evaluate phase II studies. Clin Pharmacol Ther. 2009;86(2):136-8.
FDA Conclusions
FDA Advisory Committee for Pharmaceutical Sciences and Clinical Pharmacology Meeting, March 18–19,
2008. http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4351b1-01-FDA.pdf
Clinical Response
Table 2. Summary of clinical response
Response
Arm A, 750mg/m2 over
75mins (n= 38)
Arm B, 1000mg/m2
over 30mins (n=38)
Partial Response Rate (%)
34
42
Median Survival (days)
212
287
1-year Survival Rate (%)
31.6
35.6
Conclusion – no significant difference between treatments
Toxicity
Table 3. Hematologic toxicities expressed as percentage of patients
Arm A, 750mg/m2
over 75min
Arm B, 1000mg/m2
over 30min
Toxicity
Grade 3-4
Grade 3-4
p values
Anemia
31%
33%
1.0
Neutropenia
68%
75%
0.61
Thrombocytopenia
69%
52%
0.10
Conclusion – no significant difference between treatments
Which Exposure Metric?
Structural Model Description
Objective Function Values
2 Compartments - KPD, Turnover
Emax
AUCgemcitabine
521.2
AUCdFdU
524.0
AUCdFdCTP
522.0
Dosegemcitabine
520.8
Sigmoid Emax
AUCgemcitabine
522.0
AUCdFdU
523.4
AUCdFdCTP
522.9
Dosegemcitabine
532.2
AUCgemcitabine
AUCdFdU
AUCdFdCTP
Dosegemcitabine
gemcitabine
dFdU
dFdCTP
gemcitabine
AUC-driven model
AUC-driven model
AUC-driven model
Dose-driven model
Tumour Size – Effect of Gemcitabine
Visual Predictive Check
Scatterplot Comparison
Median and 95% Interval Comparison
100
cm
cm
100
10
1
10
1
0
4
8
12
16
20
24
28
32
36
40
44
48
52
0
4
8
Week
Hi
Median
12
16
Week
Lo
Size
Hi
Median
Lo
ObsHi
ObsMedian
ObsLo
Tumour Size
Gemcitabine Dose
Tumour Size
Gemcitabine Dose-Response
Tumour Size
Gemcitabine Dose-Response
Tumour Size
Gemcitabine Dose-Response
Tumour Size
Gemcitabine Dose-Response
Tumour Size
Gemcitabine Dose-Response
Visual Predictive Check
FDA Advisory Committee for Pharmaceutical Sciences and Clinical Pharmacology Meeting, March 18–19,
2008. http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4351b1-01-FDA.pdf