Pharmacometrics Deepdive

Characterize PK within and across studies; examine different sampling and treatment regimens; characterize differences among patients

Problem

• Need to characterize kinetics of new therapies, understand if drugs are getting to the action site, in the right amount
• What are the effects of within- and between-patient variability
• What are relevant patient covariates? Absorption or effect delays?

Solution

• Pharmacokinetic analysis, by compartmental methods. Non-linear mixed effects analysis to characterize drug concentration vs. time, impact of covariates, alternative formulations, etc.
• Regulatory-ready formal reports to codify the search for and findings of the best model

Benefits

• Understand the sources of variability, including patient demographics, which are informative for trial design
• Useful for dialog with regulatory agencies to support dosing, labeling, etc.​
• Ability to leverage this for additional down-stream analysis

What is the relationship between exposure to the drug and its effects?

Problem

• Can we robustly predict our drug’s effects before we invest millions of $ in further development?
• What will our drug’s efficacy and tolerability be at doses we haven’t studied yet?
• How do exposure, patient variability and demographics affect them?
• Can we learn from other trials’ data?

Solution

• Simultaneous or stepwise analysis to characterize the dose-exposure-response relationship, incorporating necessary mechanistic considerations
• Technical competence to use the right tool for the job (NONMEM, S-PLUS/R, WinBUGS)

Benefits

• Pool data across trials to learn about covariate effects and characterize variability across trials
• Predict dose-response, with probability of success in user-friendly graphical forms that communicate to broad audiences
• Ability to leverage this for more strategic analysis

Characterize your drug exposure with minimal prior assumptions

Problem

• Need to characterize drug exposure, dose-linearity, bioequivalence ​ of alternative formulations​
• Time-sensitive summaries in clinical or preclinical setting

Solution

• Non-compartmental analysis of pharmacokinetic data​
• Regulatory-ready formal reports summarizing PK data​
• Provide results output in PP domain CDISC SDTM format ​ suitable for inclusion in clinical database

Benefits

• Get “first-cut” information on drug exposure, clearance​
• Useful summary of raw data, comparison among doses, etc.​
• Ability to leverage this for additional down-stream analysis

Drug-disease modeling combines two sets of modeling approaches: population-based models, which are typically classified as pharmacometric (PMX) models and systems dynamics models, which encompass a range of models of physiology, signaling pathways in biology, and substance distribution in the body which together are often called quantitative systems pharmacology models (QSP). Drug-disease modeling integrates PMX and QSP to include selected mechanistic aspects in a fit-for-purpose configuration to address variability and the testing of covariates. These models can be used from preclinical through Phase 3.

Problem

• What is the expected outcome of the next trial (phase II or III) given the prior information on the drug that was captured in a predictive model?
• Which trial provides the greatest value to the program?
• Which of several alternative trial designs are the best?

Solution

• Leverage existing modeling work: Predictive PK/PD modeling, meta-analysis, etc.
• Simulate model-based trial outcomes across various trial designs (dosing schemes, patient population, sample size, etc.) and keep track of how they fare

Benefits

• Robustly predict the next trial’s outcome
• Optimize probability of trial success, decrease time to filing, enhance information gained per trial, and maximize trial benefit/cost ratio
• Inform investment decision with key metrics of probability, time, cost

How do we optimize trial design to get the best probability of success at the least cost? “War game” the alternatives before investing.

Problem

• What is the expected outcome of the next trial (phase II or III) given the prior information on the drug that was captured in a predictive model?
• Which trial provides the greatest value to the program?
• Which of several alternative trial designs are the best?

Solution

• Leverage existing modeling work: Predictive PK/PD modeling, meta-analysis, etc.
• Simulate model-based trial outcomes across various trial designs (dosing schemes, patient population, sample size, etc.) and keep track of how they fare

Benefits

• Robustly predict the next trial’s outcome
• Optimize probability of trial success, decrease time to filing, enhance information gained per trial, and maximize trial benefit/cost ratio
• Inform investment decision with key metrics of probability, time, cost

Where does our drug stand vis-à-vis competitors in a crowded market? At what dose does ours fare the best?

Problem

• Does our drug stand a chance against the Standard of Care?
• What dose of my drug is superior/non-inferior to SoC?
• What are expected outcomes in various patient populations?

Solution

• Integrate public domain competitor/placebo summary statistical data into a model
• Combine public summary-level and internal patient-level data to learn from both

Benefits

• Quantitative imprint of the entire competitive market, ready for multi-faceted trade-off assessments​
• Efficacy / safety ratio comparisons across marketed and development drugs​
• Head-to-head comparisons in patient populations that cannot be studied (because of time and cost) by a single sponsor

What is the best clinical plan given the probability of success, time to market, and commercial potential?

Problem

• What is the value of information for alternative Phase II trial designs? Some delay launch for a valuable indication, but also lower the risk of expensive Phase III failure, while others do the opposite!
• Which option provides the best shareholder value, and has the best business justification?

Solution

• Employ Decision Analysis techniques to quantify the costs and risks, and characterize the net value to the company of each trial alternative
• Integrate robust probabilities generated by M&S with Decision Analysis techniques such as decision trees and other valuation methods

Benefits

• Science-driven business decisions, linking data to the recommendation via pharmacometrics
• M&S-based analysis results that “speak the language” of executives​

Given the ambiguous trade-offs inherent between efficacy and tolerability, what dose provides the best net patient benefit?

Problem

• Where do we focus to improve the drug?
• What is more important to patients and providers: more efficacy or better tolerability/safety?

Solution

• Employ the Decision Analysis technique of multi-attribute utility analysis (a.k.a. Clinical Utility) to quantify trade-offs
• Integrate with the results from PK/PD modeling to generate predictions of Clinical Utility, its uncertainty, and key sensitivities
• Includes the flexibility of utilizing market research data where available

Benefits

• Illustrate how patient net benefit is optimized and at which doses
• Dig into the results to understand what drives value, uncertainty, and thus where to focus market research, new formulations, or new therapies