Dose-Exposure-Response Analyses in MCP-Mod Framework Collaboration - - PowerPoint PPT Presentation

Dose-Exposure-Response Analyses in MCP-Mod Framework

Collaboration of Pharmacometrics and Statistics

Qiqi Deng (Boehring er ingelheim)

Acknowledge

• This is a joint work with
• Benjamin Weber (Director, Pharmacometrics, Boehringer Ingelheim)
• Other team members includes
• Dooti Roy (Biostatistics, Boehringer Ingelheim)
• Sree Kurup (summer intern, Pharmacometrics, Boehringer Ingelheim)
• Junxian Geng (summer intern, Biostatistics, Boehringer Ingelheim)

PMx/Weber - Internal use only 2

Basic Principle of Clinical Pharmacology Orally Administered Systemically Acting Drugs

Dose or dosing regimen Drug concentration in plasma Drug concentration at the site of action Effect

PMx/Weber - Internal use only 3

Dose-Response Exposure-Response Population Pharmacokinetics (POPPK)

Why exposure-response analysis may improve the modeling

• Under assumption of “exposure drives response”, PK data can help

separating PD (Pharmacodynamics) variability from PK (Pharmacokinetics).

• Berges and Chen (2013) compared exposure-response approach and

dose-response approach, and showed that

– While the accuracy for ED50 estimation was comparably good with both approaches, the precision was up to 90 % higher with concentration-response approach – The difference was most notable when clearance was highly variable between subjects and the top dose was relatively low. – The higher precision by the concentration-response analysis may lead to up to 20 % higher success rate for selecting right dose for subsequent confirmatory trial.

07/06/2017 4

MCPMod – deal with model uncertainty

Recommendations: At least 10 fold dose range, 4-7 doses, 3-5 dose response shapes

5

Model selection/average in MCPMod on dose response

• MCP step provided a way for models selection or model averaging.
• Improve performance of prediction by avoid over-fitting.

07/06/2017 6

x y

sigEmax x y

exponential

50 100 150 200 2 4 6 8 x y

Case Study Phase II Dose Finding Trial

• Six dosing groups – five active treatment groups and one placebo group
• 10, 25, 50, 75, and 100 mg
• Duration: 28 days
• Number of subjects: 40 per dosing group
• Biomarker defined on an 0-100 scale and measured at day 7, 14, 21,

and 28

• Primary endpoint: Change from baseline in biomarker at day 28
• Plasma samples at days 14 and 28 (4 samples per day per subject)

PMx/Weber - Internal use only 7

Pharmacokinetics (PK)

PMx/Weber - Internal use only 8

Cmax Cmax, steady stat Ctrough, steady state Ctrough AUC0-24,steady state

Simulation parameters

Dose – Exposure – Response

• Exposure: 𝐷𝑛𝑛𝑛 𝑈𝑈𝑈 =

𝜄𝑙𝑙∙ 𝐺 ∙𝐸𝐸𝐸𝐸 𝜄𝑊𝑒 ∙ 𝜄𝑙𝑙 − 𝜄𝜄𝜄 𝑓−𝜄𝐷𝐷

𝜄𝑊𝑒 ∙ 𝑈𝑙𝑈 − 𝑓− 𝜄𝑙𝑙 ∙𝑈𝑙𝑈

• Where 𝐵𝐵𝐷𝐵𝐵 𝑇𝑇 = 𝐸𝑙𝐸𝐸𝐸 𝑒𝐸𝐸𝐸

𝜄𝐷𝐷∙ 𝐸𝜃𝐷𝐷

• Effect = 𝜄𝐹𝐹 +

𝜄𝑓𝑓𝑓𝑓 ∙ (𝐵𝐵𝜄𝐵𝐵𝐵𝐵) 𝜄𝐼𝐼𝐼𝐼 𝜄𝐹𝐵𝐵𝐷50

𝜄𝐼𝐼𝐼𝐼+(𝐵𝐵𝜄𝐵𝐵𝐵𝐵) 𝜄𝐼𝐼𝐼𝐼 + 𝜏

Simulation parameter Value 𝜄𝐹𝐹 𝜄𝐸𝑓𝑙𝑓

100

𝜄𝐼𝐸𝐸𝐸

1.45

𝜄𝐹𝐵𝐵𝜄5𝐹

5

𝜏

25 𝜄CL (L/h) 25 𝜄Vd (L) 45 𝜄ka (1/h) 0.65

Dose

9

Exposure Response of primary endpoint at day 28

PMx/Weber - Internal use only 10

Exposure-response model could be fit via MCPMod approach to determine shape

Fitted Shapes

Exposure response modeling Dose response modeling using AUC24

11

D-R curve vs C-R curve

07/06/2017 12

x y

sigEmax x y

exponential

50 100 150 200 2 4 6 8 x y x y

• 5

5 10 50 100 150 200 250

sigEmax x y

• 5

5 10 50 100 150 200 250

exponential

Potential application for exposure-response analysis

• Assuming “exposure drives response”, it can enable clinical

trial simulation for objectively evaluating untested scenarios.

• Predication of effect at different time point post-treatment

– Assumption about delayed effect

• Extrapolation/bridging into special populations

– pediatric, renal-impaired, co-medicated, etc

• Prediction on the untested scenario are based on different

assumptions, and is yet to be tested by real data.

• Key aspects of regulatory reviews
• Question based clinical pharmacology review (FDA)

07/06/2017 13

Reference

• Berges and Chen (2013) Dose finding by concentration-response versus

dose-response: a simulation-based comparison. Eur J Clin Pharmacol.

• “Bornkamp, Björn, et al. "Innovative approaches for designing and

analyzing adaptive dose-ranging trials." Journal of biopharmaceutical statistics 17.6 (2007): 965-995.”

• Bretz, Frank, José C. Pinheiro, and Michael Branson. "Combining

Multiple Comparisons and Modeling Techniques in Dose‐Response Studies." Biometrics 61.3 (2005): 738-748.

07/06/2017 14

Knowledge sharing

07/06/2017 15