paving the critical path final japanjcop/h20/download_first_congress...supporting "free market...

© Pharsight Corporation All Rights ReservedReference ID or Date

Paving the Critical Path

Japanese Conference on Pharmacometrics

Tokyo, November 15, 2008Daniel Weiner, PhDChief Technical Officer and Sr. [email protected]

Utilization of Model Based Drug Development for Improving the

Efficiency of Clinical Drug Development

© Pharsight Corporation All Rights Reserved

Agenda

•The Business Case for M&S – Status of the Pharmaceutical Industry

•FDA’s evolving view of M&S and the Critical Path

•What is Model Based Drug Development (MBDD)?

•Case Studies Illustrating the Value of MBDD in All Phases of Drug Development

•The Status of MBDD in the U.S., Japan, Korea and China

•The Future of MBDD in Drug Development


Status of the Pharmaceutical Industry


FDA White Paper“Innovation/Stagnation: Challenge and Opportunity on the Critical Path to New Medical Products”

Beginning in 2000, a slowdown in new drug and biologic submissions to regulatory agencies occurred worldwide, partially due to increasing costs and complexity of drug development

“The vast majority of drugs in clinical trials fail. A drug entering Phase I in 2000 has no better chance to reach market than one entering Phase I 15 years ago.”

“The main causes of failure in the clinic include safety problems and lack of effectiveness: inability to predict these failures beforehuman testing or early in clinical trials dramatically escalates costs.”

http://www.fda.gov/oc/initiatives/criticalpath/whitepaper.pdf


Presented at Kitasato-Harvard Conference, Sept. 2008


Only 17 New Drugs Were Approved by the FDA in 2006


Only 17 New Drugs Were Approved by the FDA in 2007

FDA caution, 'research drought' cuts drug approvals

• By Julie Schmit, USA TODAYFederal approvals of new drugs last year sank to the lowest in five years, a drop some industry analysts attribute to more cautious regulators and less innovation by drugmakers.

The Food and Drug Administration last year approved 17 new molecular entities —active ingredients that weren't marketed in the USA before.The FDA also approved 65 original new drug applications last year, the fewest since 1999, according to data from BioMed Tracker of Sagient Research Systems.Drug applications include new molecular entities as well as new formulations or new manufacturers of existing drugs. The FDA approved 64% of the applications it decided last year, down from 73% the year before, BioMed data show.

…Ira Loss of research firm Washington Analysis says the industry's "research drought" has led to weaker applications, thus fewer approvals. An explosion of drug discoveries started in the late 1980s, peaked in the mid-1990s, then "petered out."


Even for those drugs that were approved …

Only 3 out of 10 drugs produce revenues that match or exceed R&D costsa

Only about 1 out of every 5000 molecules gets approved. If you do the math, we should be targeting identification of and eliminating the “losers” as soon as possible! But most reward systems go the opposite way!

a) Source: PhRMA 2003 Industry Profile, March 2003


Top 5 layoffs of 2007

October 15, 2007

By Maureen Martino

2007 has been a rough year for a number of the pharmaceutical and biotech industry’s biggest players. Concerns about patent expirations, falling sales due to drug safety concerns, redundancy from acquisitions and a general need streamline operations contributed to these companies’ decisions to cut employees. Check out this list of the top five pharma and biotech layoffs of 2007 for more on the cuts and a look at what these companies are doing to turn things around.

1. Pfizer - 10,000 jobs

2. AstraZeneca - 7,600 jobs (another 550 also announced)

3. Bayer - 6,100 jobs

4. Johnson & Johnson - 5,000 jobs

5. Amgen - 2,600 jobs

6. GlaxoSmithKline - 5,000 jobs*

7. Bristol-Myers Squibb - 4,800 jobs*

8. Novartis - 3,750 jobs*

*Note: These job cut announcements occurred after the date this report was originally published. GSK

announced cuts on 10/25/07, BMS on 12/6/07 and Novartis on 12/13/07.

Many companies have announced additional layoffs in 2008.

The Increase in Drug Dev. Costs, Coupled with a Decline in Approvals is Impacting Industry


The Cost Cutting Continues …

Big Pharma Layoffs Mounting -- 26 More Plants to Close, an Industrial Info News Alert

Tuesday July 22, 8:26 am ET

SUGAR LAND, TX--(MARKET WIRE)--Jul 22, 2008 -- Reported by Annette Kreuger, Industrial Info Resources (Sugar Land, Texas)

Despite the fact that millions of people consume prescription drug products on a daily basis, a tsunami of job cuts continues to roll across the Pharmaceuticals Industry. A recent analysis of Industrial Info Resources' Pharmaceutical Tracker - Online Database reveals 26 North American plants and research facilities currently targeted for closure. Pharma giant Pfizer (PFE - News) (New York, New York) has announced layoffs in Michigan, and Schering-Plough (SGP - News) (Kenilworth, New Jersey) and Johnson & Johnson (JNJ - News) (New Brunswick) are both expected to shed jobs in New Jersey.


The Cost Cutting Continues … Pfizer

Pfizer retooling, pushing emerging market sales

Mon Sep 22, 4:54 PM ET

TRENTON, N.J. - . is dramatically stepping up sales efforts in emerging markets, overhauling U.S. business operations and slashing more costs ahead of the 2011 patent loss for cholesterol blockbuster Lipitor, its head of pharmaceutical operations said Monday. Ian Read, speaking at the 2008 in New York, told analysts and investors the company has reduced annual costs by $1.2 billion from 2006 levels and expects to meet its goal of cutting a total of $2 billion by year's end. Most of the remaining reductions will come in the fourth quarter, he said, possibly signaling more job cuts.

New York-based Pfizer faces a big hurdle in replacing its nearly $13 billion in annual revenue from Lipitor, the world's top-selling drug, which loses patent protection and will face generic competition in the United States starting in late 2011.


The Cost Cutting Continues … (Pfizer cont.)

FiercePharmaOctober 14, 2008

Hinting at layoffs, Pfizer extends severance

By Tracy Staton

One hint that Pfizer layoffs will only continue: The company extended its severance-package program, which was due to expire at year's end, through mid-2009, because the drugmaker couldn't treat new layoffs differently from those cut previously. And according to sources close to the action, those layoffs probably would include a large number of sales reps.

Meanwhile, Pfizer is also considering spinning off some of the research units that it put on the "eliminate" list two weeks ago, Pharmalot reports. Plus, the company might buy a "brand-name biotech." As you know, almost every time a biotech appears to be on theblock, Pfizer is rumored to be among the pharma firms kicking tires. Which target might actually prove out is a mystery, though the gossip says Pfizer might take another look at Biogen Idec.

A new round of job cuts would add to the 10,000 Pfizer announced in January 2007 as part of a major restructuring program that also included plant closures and research-center shutdowns.


Price Controls are on the Horizon …

FiercePharmaOctober 24, 2008

Obama plan could whack Big Pharma

By Tracy Staton

Big Pharma may have been buttering up the Democrats lately with big campaign contributions, but Barack Obama's healthcare plan looks primed to hit drugmakers. According to Boston Consulting Group, Obama's plan to let the feds negotiate Medicare drug prices could take a big bite out of industry revenues. A $10 billion to $30 billion bite, to be exact.

But, according to the Wall Street Journal Health Blog, that loss could be offset at least partly by the Democratic candidate's efforts to cut the number of uninsured people. If his plans come to fruition, more people will be able to afford prescription drugs to begin with.

On the other hand, Obama's move toward negotiated Medicare drug prices could start a domino effect, inspiring private insurers to demand price cuts, too.

John McCain doesn't mention whether he'd empower Medicare to negotiate drug prices. But he does favor saving money via re-importation of drugs. And of course he's on the record with his characterization of Big Pharma as "big bad guys."


The Cost Cutting Continues …GSK

Wednesday, October 1, 2008

The Wall Street Journal HEALTH

Glaxo Plans to Cut R&D Jobs

By JEANNE WHALEN

OCTOBER 1, 2008

Drug maker GlaxoSmithKline PLC Tuesday said it plans to cut as many as 850 jobs in research

and development in the U.S. and Britain, the latest cuts in an industry grappling with declining

profits and productivity.

The cuts, about 6% of total R&D staff, would come on top of the 350 R&D jobs Glaxo said it was

eliminating earlier this year, said Claire Brough, a spokeswoman for the Brentford, U.K.,

company…


Just Released

Merck to cut 7,200 jobs October 22, 2008

By Maureen Martino at FierceBiotech

Merck said today that it will cuts 7,200 jobs--more than 12 percent of its workforce--following the announcement of a 28 percent drop in Q3 earnings. The job cuts, which will include 6,800 active employees and 400 vacancies, will be made in all areas of the company. Forty percent of the cuts will be in the U.S. and 25 percent will be senior and mid-level execs. Back in 2005, Merck cut 10,400 positions; these new cuts come in addition to that, and are expected to be completed by 2011.


Profitability Issues Plague Non-U.S. Companies as Well

Japan's biggest pharmaceutical companies report net-profit declines.

The Wall Street Journal (8/1, Shimamura) reports that Daiichi Sankyo, Eisai, and Takeda Pharmaceutical, "three of Japan's biggest pharmaceutical companies, reported net-profit declines for the April-June quarter, as they sought new sources of profits through major acquisitions and drug research."


Many Pharma’s are investing more in Biotech drugs. Likely a necessary strategy but will result in

even more layoffs.


Eventually There Will Be Price Controls in the U.S.

Total annual healthcare spending in the United States topped the $2 trillion mark in 2005, according to Center for

Medicare and Medicaid Services (CMS), and is

projected to exceed $3 trillion in 2011. That’s more

than the gross domestic product of every country in the world except the U.S., Germany, Japan, China and

Great Britain

Price controls on prescription drugs in the U.S. will put

pressure on Pharma profits

Life Science Trends ‘08

Carlyle & Conlan (C) Copyright (1-08) All Rights Reserved


The Battle Has Started

Today's Top Stories (Nov. 15, 2008)

1. PhRMA poised to fight Obama plans

By Tracy Staton at FierceBiotech

Look out, Barack Obama. Big Pharma is preparing to launch its first blitz against your healthcare plans.

The country's biggest pharma lobbying group--PhRMA--is readying a major public relations campaign supporting "free market health care." The multimillion-dollar campaign includes national television advertising, scheduled to air for the first time next week. It's all designed to head off President-elect Obama's expected campaign for Medicare to negotiate prices for drugs used by its beneficiaries.

Lots of companies stand to lose if the new administration really does reform healthcare. But drugmakersmay be closest to the line of fire. According to the Washington Times, pharma could see a $30 billion reduction in revenues if the government starts to negotiate drug prices as many other countries do.

PhRMA's push won't directly attack Obama's plans; with such a popular president-elect, such an attack could backfire. Instead, the association's ads will take a more oblique approach: "We're going to do an ad campaign designed to make people aware of the importance of preserving your free-market health care system," PhRMA's Ken Johnson told the Times. Coming soon to a TV screen near you.

- read the Washington Times article


Pharma Focus is Changing

Pharmecutical R&D Experiencing Fundamental Shift

by: Research Recap posted on: May 13, 2008

Big Pharma’s failure to develop enough new blockbuster drugs is combining with other forces to drive fundamental changes in pharmaceutical research and

development, according to Deloitte. Chief among these will be the emergence of “NewPharmaCos” concentrating on genotyped market

segments. The companies may be startups or spinoffs of existing pharmacompanies, Deloitte says in a new report The Changing Face of R&D in the

Future Pharmaceutical Landscape.

According to the report, current R&D programs focused on developing a small portfolio of high revenue blockbusters will evolve to R&D programs focused on high efficacy treatments developed for smaller patient populations based on specific genotypes. Such treatments, and their higher demonstrated efficacy,

may ultimately have the potential for higher per treatment revenues.


The Lack of Consistent Success is Adversely Affecting Drug Prices

Pharmaceutical Index (^DRG)

This capitalization-weighted index is designed to represent a cross-section of widely held, highly capitalized companies involved in

various phases of the pharmaceutical industry.


Two Cases in Point Pfizer and Merck

-27%

-43%


Unless the Industry Changes, Things are Just Going to get Worse

Big Pharma Faces Grim Prognosis

Industry Fails to FindNew Drugs to ReplaceWonders Like Lipitor

By BARBARA MARTINEZ and JACOB GOLDSTEINDecember 6, 2007; Page A1

Over the next few years, the pharmaceutical business will hit a wall.

Some of the top-selling drugs in industry history will become history as patent protections expire, allowing generics to rush in at much-lower prices. Generic competition is expected to wipe $67 billion from top companies' annual U.S. sales between 2007 and 2012 as more than three dozen drugs lose patent protection. That is roughly half of the companies' combined 2007 U.S. sales.

Starting in 2010, the pharmaceutical industry faces one of the biggest waves of patent expirations.

At the same time, the industry's science engine has stalled. The century-old approach of finding chemicals to treat diseases is producing fewer and fewer drugs. Especially lacking are new

blockbusters to replace old ones like Lipitor, Plavix and Zyprexa.

The coming sales decline may signal the end of a once-revered way of doing business. "I think the industry is doomed if we don't change," says Sidney Taurel, chairman of Eli Lilly & Co. Just yesterday, Bristol-Myers Squibb Co. announced plans to cut 10% of its work

force, or about 4,300 jobs, and close or sell about half of its 27 manufacturing plants by 2010.

Ref: The Wall Street Journal Online, Dec. 6, 2007


FDA Is Also Getting Tougher on Safety Issues – At a Minimum This Will Delay Approval Times


Pharmaceutical R&D Experiencing Fundamental Shiftby: Research Recap posted on: May 13, 2008

Discusses Deloitte report: The Changing Face of R&D in the Future Pharmaceutical Landscape.

…

In a classic Catch-22 situation, the more effective companies were at developing such mass market blockbusters, the more effective they needed to be in developing even larger revenue replacements, as patents on existing products expired and markets were lost to generics.

…

Technological advances are increasingly providing companies the opportunity to link treatments and efficacy to genetically homogeneous patient groups.

While typical blockbuster efficacy rates range from 35 - 75 percent, genotyped market segments and biomarker-assessed responses promise treatments with significantly higher efficacy rates, possibly approaching 100 percent.


FDA’s evolving view of M&S and the Critical Path


FDA View of M&S

“Clinical trial simulation—using in silico modeling—can predict efficient designs for development programs that reduce the number of trials and patients, improve decisions on dosing, and increase informativeness.”3

“In most cases, however, it is important to develop information on population exposure-response relationships… and whether exposure can be adjusted for various subsets of the population.”4

“Exposure-response data thus can add to the weight of evidence of an acceptable risk/benefit relationship and support approval.”4

3FDA Critical Path Initiative, page L-124FDA Guidance for Industry Exposure-Response Relationships


FDA is Struggling to Fund the Critical Path Initiative

The Subcommittee concluded that science at the FDA is in a precarious position: the Agency suffers from serious scientific deficiencies and is not positioned to meet current or emerging regulatory responsibilities.

1.2 Major Findings

The Subcommittee found substantial weaknesses across the Agency, with the possible exception of some drug and medical device review functions funded by industry user fees. There are several areas of greatest concern, however, which form the basis for this report’s most significant findings.

1.2.1 The FDA cannot fulfill its mission because its scientific base has eroded and its scientific organizational structure is weak.


FDA Amendments Act of 2007

On September 27, President George W. Bush signed into law H.R. 3580, the Food and Drug Administration Amendments Act of 2007, with (from left to right) HHS Secretary Michael Leavitt, FDA Commissioner Andrew von Eschenbach, and Rep. Joe Barton of Texas in the Oval Office. This new law represents a very significant addition to FDA authority. Among the many components of the law, the Prescription Drug User Fee Act (PDUFA) and the Medical Device User Fee and Modernization Act (MDUFMA) have been reauthorized and expanded. These programs will ensure that FDA staff have the additional resources needed to conduct the complex and comprehensive reviews necessary to new drugs and devices.

Two other important laws were reauthorized: the Best Pharmaceuticals for Children Act (BPCA) and the Pediatric Research Equity Act (PREA). Both of these are designed to encourage more research into, and more development of, treatments for children.

Overall, this new law will provide significant benefits for those who develop medical products, and for those who use them.


FDA hires 1,300 new doctors and scientists Thu Sep 11, 4:14 PM ET

Thu Sep 11, 4:14 PM ET

WASHINGTON - The said Thursday it has hired more than 1,300 professional staffers in a move that officials hope will help the beleaguered agency better protect the public health amid rapid technological and scientific change. "Every pay period, we have had more than 100 people walking through our doors," said Kimberly Holden, the directing the special recruitment initiative. "We have had some people who left to go into industry and ended up wanting to come back. The swings this way every once in a while."

The staffing drive, launched just five months ago, will result in an estimated 10 percent increase in the FDA's work force. Holden said the new hires will provide critical expertise after years of losing valuable medical and scientific people who took industry jobs or went into retirement. The exodus came as the agency struggled to cope with a string of drug and that damaged its reputation…

The FDA's budget is about $2.2 billion a year, with some $1.7 billion coming from taxpayers and the rest from industry user fees…


Newer FDA Initiatives

Ref: Bob Powell, Presentation to Mario Negri Institute, June 13, 2008, and Presentation at Duke University, June 23, 2008 (used with permission)


FDA Efforts to drug development uncertaintyFDA Efforts to drug development uncertainty(Critical Path)

• Improve & predict cardiac toxicity detection

Cardiac Safety Consortium (QTc Torsades)

• ↓ uncertainty in clin trial planning & failure

• ↑ Information sharing

Develop disease models, clinical trial simulation & new trial designs

• Predict drug induced nephrotoxicitypreIND

• Understand toxicity site

Nephrotoxicity Biomarkers

(Predictive Safety Testing Consortium)

• Predict and avoid adverse reaction

• Improve benefit-risk for new drugs

Serious adverse drug events (SJ Syndrome)

• Better efficacy definition

• Early go/no go

• Tool linking pre-NDA to clinical use

Oncology Biomarker Qualification Initiative

●Non Hodgkin’s Lymphoma●Non-small cell lung CA

Potential ImpactProject


NephrotoxicityNephrotoxicity BiomarkersBiomarkers

Current markers inadequate (e.g., creatinine)● Delayed detection● Not specific to damage location● Not mechanistic

New biomarkers available● Preclinical to clinical development evolution● Gain mechanistic knowledge● May demonstrate species differences● Detect toxicity earlier


Disease ModelsDisease Models(trial design & endpoints)(trial design & endpoints)

Objectives● Use prior data plus statistical analysis & simulation to solve

regulatory problems● Share solution + models of prior data publicly

Collaboration: Clinical (OND), Biostatistics (OB), OCPProjects● Parkinson’s disease: trial design to detect disease progression

change http://www.fda.gov/ohrms/dockets/ac/cder06.html#PharmScience

• Critical to understand disease/baseline characteristics, diseaseprogression, placebo/drug effects, and statistical issues (Missing data, etc)

● Non-small cell lung cancer: predictive value in 2D imaging for disease progression-8 NDAs

● Alzheimer’s Disease: trial design to detect disease progression change. Funding by AAPS

● …..


Computing Platform (Large storage, Fast network, Fast Processors)

Scientific Computing Support Services including Collaboration

• Critical Path success

• ↑ Development efficiency-productivity

• ↑ Communication clarity to patients

• Janus implementation

• Targeted exploratory analysis

• Modeling & simulation driven trial design

• New tool qualification

Regulatory Science Research

• Biomarker qualification

• Approval endpoints

• Clinical trial design

• Benefit-Risk patient communication tools

• ↑ Productivity-Quality

• ↓ Time-resource

• Move resource to other ↑ priority topics

• Janus implementation

• Best practice → Tools

● Efficacy-Safety● Risk-Benefit

Regulatory Decisions

• ↑ Speed, Quality

• ↑ Protect public health

• ↑ Confidence

●Inside●Outside

• Automated information mgmt (LIMS, Janus)

• Safety signal detection tools

• Best practice → Tools

Health Crisis

(Melamine, Vioxx)

Anticipated ImpactMeans to AddressChallenges/Mandate

FDAPlanned FDA IT Initiatives


Quantitative Benefit-Risk Path

Current State

Future State

Path Foundations

• Personalized Medicine

• Implement FDAAA changes

• Implement FDA modern bioinformatics tools

• Expand worked examples of various tools (FDA, industry, academics) over life-cycle…e.g.,

• Patient stated preference

• Decision analysis

• Establish routine public communication venues including regulatory applications

FDAAA is the FDA Amendments Act of 2007


Presented at Kitasato-Harvard Conference, Sept. 2008


These Models are Available to the Public

Pharsight Uses FDA Disease Model to Support Oncology Drug Development

Use of FDA Model in Non-Small Cell Lung Cancer Expected to Inform

Quantitative Predictions for Experimental Cancer Treatment

MOUNTAIN VIEW, Calif., November 20, 2007 – Pharsight Corporation (OTC BulletinBoard: PHRS), a leading provider of software, strategic

consulting, and regulatory services for optimizing clinical drugdevelopment, today announced that its Strategic Consulting Services

Group is using a quantitative disease model developed by the Food and Drug Administration (FDA) for non-small cell lung cancer (NSCLC) to

support an oncology drug.

…


What is Model Based Drug Development (MBDD)?

Daniel Weiner, PhDChief Technical Officer and Sr. [email protected]


A Distribution of Trial Outcomes is Associated With Every Protocol

Our Dilemma is That We Only Observe One of the Possible Outcomes

How do we deal with this? By building a virtual model of the trial and simulating the range of outcomes!

Range of possible outcomes


Model Based Drug Development

FDA’s characterization: the development and application of pharmaco-statistical models of drug efficacy and safety from preclinical and available clinical data to improve drug development knowledge management and decision making.

If implemented, this facilitates the ability to base important decisions on quantitative inputs.


A Comment …

There have been many presentations, publications, etc., showing examples of building models in various stages of drug development and calling that MBDD. But in my humble opinion that is just that, building models.

Model based drug development is a process whereby you establish decisions that need to be made and the criteria for making them.

The modeling and simulation that you do are inputs to the decision making process, and should only be done in as complex a manner as needed to support the decisions that need to be made.


The Model Based Drug Development Process

Establish decision criteria (at least identify the key decision-drivers).

Models only need to be as complex as needed to make a decision

Leverage prior knowledge from all available sources (more on this in the next lecture)● Data on the NCE of interest

• Preclinical, Phase I safety & biomarkers, clinical safety & efficacy …

● Knowledge about the target disease & affected physiologic systems

● Knowledge/data on related compounds• From proprietary or public sources• Analogs or competitors


The Model Based Drug Development Process (cont.)

Build models for all responses key to strategic decisions based on that knowledge● Disease progression

● Drug models for the drug of interest and related or potentially competing drugs

• Efficacy-related outcomes• Safety-related outcomes

● On-trial behavior, e.g., dropouts and compliance

● Market and financial models

● It is possible that you may be able to show a drug is safe and effective, yet it’s profile (expected label) or resulting market are such that it will have a poor return on investment


Some General Comments on Models

Models can be empirical or mechanistic. We will consider several examples

Models are dynamic and not static.

● They should evolve as more data and knowledge is available. ● As such, repositories are need to support them.

The “Real” value of models is not their application to data at hand; rather it is their ability to be used to predict future outcomes along with some measure of uncertainty


Models Provide a Rational Framework for Integration and Organization of Disparate Compound/Biologic Knowledge

Understanding and embracing uncertainty is key to quantitative decision making and effective risk management.

Model Based Drug Development

● Is an integrated, data-driven, model-supported decision-making methodology

● Integrates all relevant public and proprietary data spanning:• discovery to clinical• in-house data to competitors’ information• healthy volunteers to patients

● Involves building probabilistic model(s) of the compound’s attributes and product profile in the context of a competitive landscape

● Simulates development scenarios of interest


To Simulate Development Scenarios of Interest a Number of Sub-Models are Developed and Integrated

Sub-models differ in focus, mechanistic-depth, required disciplinary expertise, the quantity and quality of data they are developed from, and the extent to which they serve to summarize proprietary client data

• Preclinical data• Early-phase trial results• Late phase results• In-house experts• Public information

InformationSources

Pharmaco-kinetics

Pharmaco-therapeutics

ClinicalEvents

Efficacy & Tolerability Endpoints

Pharmaco-dynamics

FinancialValue

MarketEconomics

DosingRegimens Exposures

1 or more drugs

ClinicalProfile

Companies often stop here


Sample Model Types

Pharmacokinetic (PK)

Pharmacodynamic (PD) and (PK / PD)

Compliance

Patient demographics

Sampling time and dosing time deviations

Between and within subject variability

Site to site variability

Disease progression

Other …

Now for more detail about how MBDD works


MBDD is a Systematic and Accountable Process

Modeling Steps● Define your next drug development decisions in terms of program value.● Explore all the proprietary data you have that is relevant to your decisions.● Augment your proprietary data with all relevant public data. ● Build a mathematical, quantitative model of the complete state of knowledge

relevant to your decision based upon the full data set.● Qualify the model against your data in order that it provides you with an well

defined description of each assumption made in building your model.

Simulation Steps● Simulate outcomes resulting from various possible decision scenarios using the

final model.● Predict the outcome of any relevant decision scenarios and quantify the most

likely result.

Leverage the above in such a way that all decisions are based onquantitative inputs.


MBDD focuses on therapeutic and commercial value

Every development decision is ultimately a decision about value. The overriding question is always, “how best to spend development budget and time to get regulatory approval for the most valuable label possible?”

The Value Function:

To maximize value and reduce risk should we

• Proceed to the next phase of development ?• Stop development altogether ?• Continue to gather data in the current phase ?

To maximize value and reduce risk should we

• Proceed to the next phase of development ?• Stop development altogether ?• Continue to gather data in the current phase ?

Program Value =- Development Costs

- Time to Market Costs

+ Value of Approved Label

The Decision:


MBDD Should Employ an Iterative Learn-Confirm1Approach

Learn-Confirm is the context in which MBDD predicts whether program value will be maximized by proceeding to the next phase, cancelling the program or gathering more data in the current phase and then re-analyzing.

Learn

Decision• Proceed to Next Phase ? • Stop Development ?• Continue Current Phase ?

Decision• Proceed to Next Phase ? • Stop Development ?• Continue Current Phase ?

If prediction looks like this, Continue Current Phase or Accept Risk and Proceed

Confirm Learn Confirm

Phase I

Use MBDD to predict response & compare to Standard of Care

1Sheiner LB, "Learning VS Confirming in Clinical Drug Development," Clin. Pharmacol. Ther. 1997, 61:275-291.

Phase IIa Phase IIb Phase III

If prediction looks like this, Stop Development

Not Competitive

Response at Given Dose

Competitive If prediction looks like this, Proceed to Next Phase

Standard of Care


Workflow for CATD learn/Confirm Cycle

Model

Capture

SimulateLikelihood of Success?Optimize

ClinicalTrial

Collect data

TherapeuticKnowledge

STOP

Meta-Database data and Prior Knowledge

No

Yes


Learn and confirm in non-clinical researchPreclinical - learning

Modeling and Simulation TasksQuestions?Objectives

Predict human efficacy

Predict human clearance using IV/IV data

Assess margin of safety from target efficacy concentration & exposure data

from tox studies

Integrate clearance, potency, bioavailability for human dose projection

Design animal PK/PD experiments, analyze data

Understand MOA

Identify lead candidate based on desired attributes

Evaluation of biomarkers relative to decision criteria

Dose range in early clinical trialsAccrue tox data to support human dosing

Guide development strategy with integrated decision criteria

Safety/efficacy characteristics in animal models

Demonstrate activity in animal models


Learn and confirm in clinical researchPhase 1 - learning

Modeling and Simulation Tasks

Questions?Objectives

Assess parameters in target and sub-populations

Assess dose-concentration-response curve

Quantify variability in PK and PDPD effects in FIM studies

Simulate exposure and response for intended dosing regimens

Linear and/or predictable PK in humans

Collect initial PD data (if possible) and assess PK/PD attributes

Develop or update PK/PD modelsMTDAssess MTD


Learn and confirm in clinical researchPhase 2A (PoC) - confirming



Develop drug/disease model to understand time-course of

disease progression and dose-response to interventions

Attributes of drug in target population compared to

existing therapy

Demonstrate efficacy in the intended population


Learn and confirm in clinical researchPhase 2B - learning



Assess efficacy/toxicity profile of drug vs. comparators

Validate Pop PK/PD model

Assess impact of covariates using Pop PK/PD model

Analyze dose-exposure-response data using a model based approach

Assess probability of success for a candidate study designs

Use simulations/optimization tools to design dosing and sampling schemes

Critical aspects reflecting value of PoC vs. larger Phase II (duration, sample

collection schedule, biomarkers). Do we have adequate information to terminate

development or continue to the next phase?

Establish doses to be further evaluated in Phase III

Simulate outcomes based on design and variables

Trial design to unequivocally demonstrate efficacy in target population (dose

range/frequency)

Identify optimal use in target population


Learn and confirm in clinical researchPhase 3 - confirming



Assess need for dose adjustment in special populations

Confirm dose/exposure response-relationship in target/sub

populations

Do we have adequate information to decide whether to file, collect

addition data, or terminate?

Assess impact of covariatesDose change in a sub or special population (degree of change)?

Validate PopPK/PD modelDo the intended doses (and

regimen) demonstrate safety and efficacy in the target population?

Demonstrate safety and efficacy


Pre-Clinical Data

Safety Trial

PK-PD Example: A model summarizes the current state of knowledge

In this example, the team wants to answer the question, “do I have a competitive dose? “

In the modeling step, scientists model the dose-response of the compound and its competitors by combining proprietary data about the compound with public information such as scientific reports and competitor labels to produce an aggregate data set. Next, proven MBDD methods are used to fit a mathematical expression—a model—to the data and then quantify the quality of the fit.

Latest Phase II Trial

Scientific Literature

CompetitorLabelInformation

Model Assumptions

1.

2.

3.

Dose

Resp

onse

Previous Phase II Trial

The model summarizes and quantifies what is known.The model assumptions detail what is not known.


+C = (D/V)(Ae-αt + Be -βt)

PK-PD Model

E(= E0 + (Emax- E0) • C

C+ EC50


PK-PD Example: Simulations produce predictions of performance (plus uncertainty)

In the simulation step, the predicted response for various doses is extracted from the model and compared to the standard of care.

This is done by Monte Carlo simulation: applying different decision scenarios to the model over and over again. The outcomes from the simulation replicates are analyzed to predict the most likely result. These predictions represent new information, derived from all the existing data, that the team can use to decide whether and how to proceed with this drug program.

Model Assumptions

1.

2.

3.

Simulation Step

The model predicts that, given the scenario inputs and the assumptions, there is a competitive dose. The team can decide to confirm this prediction, that is, test the assumptions, in the next development phase.

The model predicts that, given the scenario inputs and the assumptions, there is a competitive dose. The team can decide to confirm this prediction, that is, test the assumptions, in the next development phase.

Response at Given Dose

Not Competitive

CompetitiveDecision Scenario Inputs• Dose• Population• Competitor

Prediction:

+

Standard of Care

C = (D/V)(Ae-αt + Be-βt)

PK-PD Model

E(t) = E0+ (Emax- E0) • C

C + EC50


Models can be built to support any type of decision By integrating different kinds of data, scientists can use MBDD methods to produce quantified predictions for any drug development decision, in any regulatory phase.

• PK-PD Models

• Disease Models

• Compliance Models

• Drop-out Models

Different Kinds of Models

• Trial Models

• Competitor Models

• Commercial Models

• Portfolio Models

Clinical Trial Data


Marketing Study

Portfolio Data

CorporateFinancials


Is there a clinical trial design that will show PoC and find the best dose?

Is it worth developing a new dosage form?

Should we continue this development program?

What is the optimal patient population for this drug?

What is an optimal regulatory strategy?

Note the focus is on leveraging

To optimize decisions regarding


MBDD applies across the regulatory timeline

Preclinical Phases I/IIa Phase IIb Phases III/IV

Is there a clinical trial design that will show PoC and find the best dose?

Is it worth developing a new dosage form?

Should we continue this development program?

What is the optimal patient population for this drug?

Which indication should we go into first to maximize the value of the program?

Is this treatment likely to be as good as the competition?

Should we in-license this compound?

What candidate should we take forward to human clinical trials?

What is the probable clinical dose-response in humans?

What’s the best dose and schedule?

What dose provides the best benefit/risk profile?

What is an optimal regulatory strategy?

What are additional indications?

What are the most important attributes of a 2nd generation compound?

time

How do we demonstrate efficacy?

Have we demonstrated improved benefit/risk compared to standard of care?

Sample questions by phase of development.


FDA Assessment of Potential value of MBDD by phase

Used with permission from a presentation by Dr. Bob Powell


Case Studies Illustrating the Value of MBDD


Pre-clinical (In-vitro → in-vivo)

● Dose selection for First-in-Humans (FIH) study

Early Development

● Concentration-QT modeling in early development

Late-Stage Development

● Model-based NCE profile assessment versus competing therapies

Post Approval

● Model-based pediatric dose selection for an approved biologic

Case Examples Across All Phases of Drug Development


Rationale for Investing in FIM Dose Calculation Methodologies

Investing in approaches that offer relatively small improvements in prediction of human NCE attributes from preclinical/translational data offers relatively large returns● “10% preclinical prediction improvement has the capacity to save $100

million per drug” (Expert Opin. Drug Saf., 2007. 6(2), pp217-224)

Regulatory climate is shifting towards increased focus on safety issues● TGN1412 and other experiences● New Guidance documents focused on safety

FIH calculations of likely efficacious dose has implications for the manufacturing “cost of goods”

Wagg JK. Update on Quantitative Approaches for Predicting First-in-Man (FIM) Dosing Regimens. Pharsight European Modeling and Simulation Seminar. May 30, 2008. Paris, France.


Overview of Methodologies

1. Allometric scaling of doses (FIH start & end doses)

2. Build PK/PD models, scale parameters to humans and simulate range of likely phase I/II/III outcomes for trial designs of interest (initial estimate of clinical profile, inform trial designs, some times FIH start doses)

3. Build semi-physiological PK/PD models, scale parameters to humans and simulate range of likely phase I/II/III outcomes for trial designs of interest (initial estimate of clinical profile, inform trial designs, some times FIH start doses)

Approaches 2 and 3 allow critical evaluation of different trial design options including adaptive trial designs and represent examples of the broader discipline of Model Based Drug Development (MBDD)


Sample Allometric Scaling Model

Allometric Prediction of Human Systemic Clearance of Xenobiotics from Animal and Human Data

In(Typical Species Body Weight)

In(C

lear

ance

, vol

/tim

e)

++

++

+

+

Expected Human Clearance

A Potential realization of Human Clearance

Source: [Hu and Hayton, AAPS PharmSci 2001; 3(4) article 29].


Alternative Physiological Recirculatory Model

Moriwaki, T., Yasui, H., and Yamamoto, A. A RecirculatoryModel with Enterohepatic Circulation by Measuring Portal and Systemic Blood Concentration Difference, J Pharmacokin PharmacodynV30(2), Apr 2003.

Sample Physiological Model


Synthesis kout · R

R

Warfarin

Res

pons

e

Time

Slope = -kout

Inhibitionby

warfarinR = Prothrombin complex

activity (PCA)

Sample Semi-Physiological Model

Vitamin Kepoxide

ReducedVitamin K

Vitamin KVitamin Kreductase

Vitamin K epoxidereductase

Warfarin

Prothrombin complexprecursors

Prothrombin complex (R)VII, IX, X, protein C, S

γ-Glutenyldecarboxylase


Allometric scaling was used to predict human pharmacokinetics.

Preclinical PK/PD data from cynomolgous monkey, relative potencyinformation and literature data was used for simulation.

A range of doses (30-fold), regimens (QD and BID) and bioavailability fractions (5 to 50%) were used in simulations to project human PK vs. response profiles. The combination of dose and bioavailability ranges was chosen to compensate for any misspecification due to projection method or underlying assumptions.

Target therapeutic range was determined using publicly availableliterature for three comparators.

Example: How can preclinical data be used to support dose selection for a FIM study?

Kastrissios H. Communicating the Value of Pharmacodynamic Modeling in Drug Development. Presented at: Association of Regulatory and Clinical Scientists(ARCS) 17th Annual Congress; May 28-30, 2008; Sydney, Australia


Comparisons across regimens and drugs showed a favorable predicted response

0

20

40

60

80

0 100 200 300

Multiple BID

0 100 200 300

Multiple QD

Dose (mg/day)

%Ti

me

in T

h. W

indo

w


The response was comparable to competitorsA

dvse

rse

Eve

nt P

roba

bilit

y (%

)

0

20

40

60

80

100

A1 A2 A3 A4 A5 A6 0 B1 B2 B3 N1 N2 N3

Drug A Dose Drug B Dose NCE Dose(mg/day) (mg/day) (mg/day)

QD regimenBID regimenDrug C

Human projections for the NCE of interest identified a dose which provided a similar safety profile to that of comparators.




Early Development




Post Approval




C-QT modeling is a valuable tool early in clinical development prior to conduct of a TQT study

Early studies (SAD and MAD) provide a chance to develop a C-QT model and inform critical development decision-making.● Identify drugs with “no-go” levels of QT prolongation faster. ● Better calculate risk-benefit and clinical window (viable dose range)● Determine when and how (or if) a TQT should be performed (only run TQT for

compounds entering later stages of development)

For situations where a TQT cannot be performed (such as oncology), pooled analysis of early studies is critical to understanding QT risk.

To enable valid modeling and prediction, these studies should bedesigned with the QT analysis in mind. (Not a large burden)


C-QT model building in early development is slightly different from what is typically done in the TQT setting

Model unadjusted QTc, not delta or delta-delta QTc.● Basing the model on “whole” QTc avoids any “regression to mean” bias.● It is important to understand the differences in baselines across

populations.● Absolute QTc is also of interest in determining pro-arrhythmic risk.

Use predicted concentrations as opposed to observed● Linked to the population pharmacokinetic model.● Clarifies how population characteristics contribute to variability in

exposure separately from variability in exposure-response.● Using predicted concentrations avoids undue influence of outliers from

measurement error.

May also include time of day directly in the model.


As circadian rhythm is present in QTc, it is important to model time of day (often using a cosine function)

Piotrovsky V (2005), “Pharmacokinetic-Pharmacodynamic Modeling in the Data Analysis and Interpretation of Drug-induced QT/QTc Prolongation.” AAPS Journal 7(3):609-624.


Here is an actual example using the cosine method

1 21 2

2 ( ) 2 ( )( ) cos cos24 12t tf t A Aπ φ π φ− −⎛ ⎞ ⎛ ⎞= +⎜ ⎟ ⎜ ⎟

⎝ ⎠ ⎝ ⎠

360

380

400

420

440

8 10 12 14

Time of the Day [hrs]

QTc

Indi

v [m

sec]

Baseline Visits


When possible, we use a mixed effects model for individual adjustment of QT values for heart rate

Individual adjustment should give more accurate answers when using the model for prediction.

QTc = QT/RRN

In this case, the typical exponent was 0.24 (as opposed to 0.33 [F] or 0.50 [B]).

We have seen this make a 20% difference in the estimate of the slope. This can be the difference between a “positive” or “negative”TQT.

320

340

360

380

400

420

440

600 800 1000 1200 1400

Bazett Fridericia

No Correction

320

340

360

380

400

420

440

600 800 1000 1200 1400

Indiv. Correction

RR [msec]

QT

[mse

c]

Baseline measurements


Steps in the pooled analysis

A general outline is as follows:

1. Correct for heart rate (typically Fridericia, plus maybe individual)

2. Start with simple mixed effects model of baseline

3. Estimate time of day model with baseline data (at individual level if data permit)

4. Explore and model relationship of QTc with concentrations

• predict concentrations for all ECG samples, with/without matching PK draw• examine linear, log-linear, and saturable models• check for hysteresis/metabolite effects, as applicable• fit models at individual level as data permit


Here is a detailed example of how C-QT helps you via the pooled analysis of early phase data

Anti-diabetic compound (in Phase 2)

Modeling and simulation had been conducted throughout clinical development for● selecting doses and regimen● illuminating efficacy/safety tradeoffs● qualifying biomarkers● modeling disease progression

QT was one of several exposure-response endpoints modeled

QTc dataset was from a single Phase I crossover study (pre-TQT)● Maximum dose was 4X planned therapeutic dose● Other studies only had ECG at pre-dose and study exit● Primary study had 24 subjects in a 3 period crossover {10, 20, 60 mg}.

• Each period: Drug X QD for 4 days • Intensive PK sampling on each Day 1 (up to 24 h post dose) and Day 4 (up to 72 h post dose).• ECGs taken on each of the 4 dosing days at pre-dose, 2, and 4 hours post-dose.• Manually-read triplicate observations at each sample point.• Each observation was the average of 3 cardiac beats.


No relationship was seen with concentration

0 20 40 60 80 100 120

10 mg

340

360

380

400

420

440

0 50 100 150 200 250

20 mg

340

360

380

400

420

440

0 200 400 600

60 mg

Predicted DRUG X Concentrations [ng/mL]

QTc

Indi

v [m

sec]


The best fit of the model confirmed this

-80

-60

-40

-20

0

20

40

60

0 200 400 600

Predicted Drug X Concentrations [ng/mL]

delta

QTc

I [m

sec]

Note: adjusted for estimated subject-specific baselines


Modeling predicted the outcome of the TQT

The predictions made by the model were useful.

● Doses and concentrations covered the supratherapeutic scenario● Predictions from the model gave reassurance regarding future TQT result

Time (h)

0 2 4 6 8 10 12 14 16 18 20 22 24-10

-5

0

5

10

15

20A - Therapeutic DoseB – Supratherapeutic DoseD - Moxifloxacin 400 mg

ddQ

TcF

(mse

c)

Plasma Concentrations (ng/mL)0 600 1200 1800 2400 3000 3600

-75

-60

-45

-30

-15

0

15

30

45

60

75

A – 60 mg B - 200 mgPlacebo

Slope = -0.004r ² = 0.004


M&S of QTc Prolongation: References for further reading

General:● Garnett CE et al. (2008) “Concentration-QT relationships play a key role in the evaluation of proarrhythmic risk

during regulatory review.” Journal of Clinical Pharmacology 48:13-18.● Bloomfield D and Krishna R (2008) “Commentary on the clinical relevance of concentration/QT relationships for new

drug candidates.” Journal of Clinical Pharmacology 48:6-8. ● Russell T et al. (2008) “A perspective on the use of concentration-QT modeling in drug development.” Journal of

Clinical Pharmacology 48:9-12.● Carrothers TJ, Khariton T, Rohatagi S (2008). Performance Characteristics of Pooled Early Phase Concentration-QTc

Modeling. 37th Annual Meeting of the American College of Clinical Pharmacology (ACCP); September 14-16, 2008; Philadelphia, Pennsylvania.

● Rohatagi S et al (2008). Role of Modeling and Simulation in Evaluating the QTc Prolongation Potential of Drugs. American Conference on Pharmacometrics (ACoP); March 9-12, 2008; Tucson, Arizona.

● Carrothers TJ (2007) “Concordance between the exposure-QTc response relationship in the pooled phase 1 data vs. thorough QTc study.” Presented at: 2nd Annual DIA Cardiac Safety Conference; December 4, 2007; Prague, Czech Republic. Available at http://pharsight.com/library/library_cs.php

Methods:● Hutmacher MM et al. (2008) “Performance characteristics for some typical QT study designs under the ICH E-14

guidance.” Journal of Clinical Pharmacology 48:215-224. ● Piotrovsky V (2005) “Pharmacokinetic-pharmacodynamic modeling in the data analysis and interpretation of drug-

induced QT/QTc prolongation.” AAPS Journal 7:609-624.● Li L et al. (2004) “QT analysis: a complex answer to a ‘simple’ problem.” Statistics in Medicine 23:2625-2643.

ICH E14:● International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for

Human Use. (2005) The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs: E14. http://www.fda.gov/cber/gdlns/iche14qtc.htm




Early Development




Post Approval




What is the product profile of an NCE versus competing therapies ?

Gemcabene (CI-1027) is a non-statin compound developed as a low-density lipoprotein cholesterol (LDL-C) lowering compound.

Based on a beneficial effect of the drug on LDL-C in several phase I and IIa trials. it was decided to initiate a study in hypercholesterolemia.

Key question: “Given the LDL-C lowering effect of gemcabene in combination with a statin compared with competing therapies, should clinical development continue?”

A second objective was to effectively communicate the critical drug attributes to the clinical team to facilitate decision-making

Hermann D, Wang W, Falcoz C, Hartman D, Mandema J. Strategies to Improve Model-Based Decision-Making During Clinical Development. Presented at: Annual Meeting of the Population Approach Group in Europe (PAGE); June 2005; Pamplona, Spain. Reprinted courtesy of PAGE.


Strategy: Efficient Model-Based Development

A Phase IIA trial was planned to assess gemcabene LDL-C lowering ability, alone and in combination with atorvastatin

To aid decision-making, the team agreed to undertake a dose-response analysis of gemcabene trials as well as statins and ezetimibe (competitor) using literature data● 21 trials were included (~10000 patients)● Statins (atorvastatin, rosuvastatin, simvastatin, pravastatin,

lovastatin)● Nonstatins (gemcabene historical data, ezetimibe, mono- and

combination therapy)● Models were built for 7 efficacy and safety endpoints that drive

decision-making, and were updated with the Phase IIA trial results


Methods: Meta-Dose-Response Analysis

Mono-therapy LDL-C % change dose-response:Statins and Non-statins: gemcabene, ezetimibe

Interaction term added to describe combinations

Weighted (by variance) non-linear mixed effects (study level random effect) regression to estimate model parameters.

nn

n

drug EDDoseEDoseE

50

max

+⋅

=

εη ++⋅⋅+++= −− statinnonstatinstatinnonstatin EEγEEEchangeLDL 0%


Results: The Model Described Mono- and Combination Dose-Response Well for Ezetimibe …

E 0 = statin alone E 10 = statin + ezetimibe 10 mg


… And Gemcabene

A 0 = gemcabene aloneA 10 = gemcabene + atorvastatin 10 mg etc.


Pre-Clinical Data

Summary of process so far ….


Model Assumptions

1.

2.

3.Safety Trials

Population Modeling

Techniques, Model

qualification



+

Dose

Resp

onse

Phase II Trials


Simulations

Dimensions of Decision Space• X Endpoints• Y Drugs/Doses• Z Covariates

X*Y*Z dimensions

e.g.,1000 simulations/dimension= [X*Y*Z]* 103 data points

C(t) = D/V(Ae-αt + Be -βt)

PK-PD Models

E(t) = E0+ (Emax- E0) • C(t)

C(t) + EC50


What is the expected clinical response for a treatment in a particular patient population?

What is the level of certainty surrounding predicted response?

How do different treatment strategies and target patient sub-populations impact response?

What is the probability that response is less or greater than a specific target?

What dose is required to achieve a target response?

What is the probability of achieving a specific efficacy target while keeping probability for adverse events below a certain level?

How do the attributes for the compound compare to competitors?

What is optimal positioning strategy versus competitors to balance safety and efficacy?

Models provide answers to important clinical questions.


Difference in LDL % change frm baseline vs CI1027Atorvastatin: 0Ref: Atorvastatin: 0 + Ezetimibe: 10

CI1027

LDL

% c

hang

e frm

bas

elin

eD

iffer

ence

in

-20

-10

0

10

20

0 200 400 600 800 1000

Superior

Equal

Inferior

Q1: What is the probability that gemcabene mono-therapy is clinically superior to ezetimibe 10 mg?

Inferior Equal Superior

900 0.0% 2.8% 97.3%

Gemcabene (mg)300 10.0% 89.9% 0.1%450 0.0% 53.5% 46.5%600 0.0% 7.2% 92.8%

Range ± 5%

Gemcabene at a dose of 600 mg or more is superior to ezetimibe


1 PlotLDL % change frm baseline vs Atorvastatin

Treatments

Ezetimibe 0Ezetimibe 10

Atorvastatin

LDL

% c

hang

e frm

bas

elin

e

-80

-60

-40

-20

0

20

0 20 40 60 80

.

.

.

1 PlotLDL % change frm baseline vs Atorvastatin

Treatments

CI1027 0CI1027 900

Atorvastatin

LDL

% c

hang

e frm

bas

elin

e

-80

-60

-40

-20

0

20

0 20 40 60 80

.

.

.

Q2: What is the probability that, in combination with a statin, gemcabene is clinically superior to ezetimibe?

Atorvastatin dose (mg)

LDL % Change from Baseline

Atorvastatin alone

+ Gemcabene 900 mg + Ezetimibe 10 mg

Atorvastatin alone

Gemcabene combination will not provide superior LDL-C lowering relative to ezetimibe


Q3: Given the magnitude of LDL-C lowering across the gemcabene + statin dose range should clinical development continue?

Data Analysis Method

Data Base Assumptions

Mean (95% CI)

Comments

Gemcabene Combo - mono

ANCOVA Phase IIA trial only (n=255)

Few -4.8 (-12.3 to 2.7)

Traditional analysis

Meta-Dose- Response

Phase IIA trial pooled with

relevant historic data

Many -2.5

(-5.8 to 1.2)

Width of CI decreased ½ compared to traditional

analysis

Ezetimibe Combo - mono

Meta-Dose-Response

Phase IIA trial pooled with

relevant historic data

Many -8.6

(-9.1 to -8.3)

Gemcabene combination has very low probability

of reaching target competitor level of LDL-

C lowering

The gemcabene CI from the meta-analysis does not overlap ezetimibe CI, clearly suggesting that

gemcabene is unlikely to lower LDL-C to the extent necessary to compete with ezetimibe.

"The results showed that the new drug was unlikely to outperformits main competitor; Pfizer discontinued development. The

modeling project supported a more confident decision without investment in additional trials and allowed team members to

re-deploy to other programs. It provided an enduring, evolving knowledge repository to support future development projects.

DMX enabled team members to evaluate complex product tradeoffs and ask targeted what-if questions in real time.”

Source: “PK/PD Simulation Speeds Decision Making.” Bio-IT World Best Practices[serial online]. January 23, 2006




Early Development




Post Approval




Pediatric Drug Development: Clinical Needs

• Current Status:● Drug dosing is highly empirical and based on trial and error● Linear extrapolation (mg/kg or mg/m2) may not be adequate● PK/PD relationship is unknown in pediatric patients

• Clinical Needs:

● To determine a safe and effective dose● A easy way to calculate individualized dose if needed● To understand the impact of developmental changes on PK/PD

Mouksassi S, Marier JF. Modeling and Simulation for Optimal Drug Developmentin Pediatric Populations. Presented at: 2nd World Conference on Magic Bullets (Ehrlich II); October 3-5, 2008; Nürnberg, Germany.


Pediatric Drug Development: Challenges

• Data collection difficulties● Invasiveness● Limited sampling possibilities sparse data● Recruiting and consent

• Costs/Ethical Issues


Pediatric Drug Development: Opportunities

• Prior and current Knowledge

● Extensive adult data available● Current understanding of physiology and pharmacology● Significant knowledge available in industry, academia and regulatory agency

need to share

• Non-linear mixed effects modeling and Simulation

● Can handle sparse and unbalanced data● Covariate analysis shed light on important determinants of

exposure/safety/efficacy● Innovative models can reduce sampling ( e.g. Time-response or PK-PD models)

• Allometry theory and developmental changes

• Optimal Pediatric studies

● Clinical trial simulations to design successful trials● Population optimal design limited sampling of high value● Adaptive Bayesian design to inform best the PK/PD relationship


Pediatric Drug Development: Efficiency is the Key

Disease Model

Studied System

Physiology

Pharmacology

Drug Model

Efficacy

Outcomes

Safety

? Developmental Changes

Is a pediatric PK/PD Safety

study required ?

If yes what is the study design that best informs the

decision ?


Case Study: Supporting Pediatric Dose Selection

Background● Biologic drug X approved in adults● PK model one-compartment, the drug is excreted by receptor mediated mechanisms● PD model steep direct Emax (on-off) model, need to keep concentration above a

threshold.● Disease progression is not likely to be different in pediatric patients● Limited prior experience demonstrated that the EC50 is similar in pediatric patients● Allometry theory can be used to predict pediatric clearance

• Objectives● To design an optimal sampling strategy for PK/PD determination in pediatric patients● To design a dosing strategy for a pediatric trial that will ensure adequate PD response

(> 95% above threshold).

• Tools● Optimal design based on the population fisher information matrix and partial

derivatives.● Various clinically feasible dosing scenarios were tested to find the dosing strategy that

has the minimum likelihood of PD response breakthrough.


Scaling from Adults to Pediatrics: Predicting PK

( )0.75

( )70

WeightCL TVCL Maturation age⎛ ⎞= ⎜ ⎟⎝ ⎠i i

Drug X is cleared by receptor mediated mechanisms

No correlation with renal function.

Maturation can reflect receptor density or disease interaction with age. No maturation function was incorporated in this CTS exercise

The allometric model was also used for volume of distribution

1

70WeightVc TVVc ⎛ ⎞= ⎜ ⎟⎝ ⎠i


Optimal Design Points Based on Partial Derivatives

0 2 4 6 8 10 12 14 16WEEK

-70

-50

-30

-10

10

300

600

-17000

-12750

-8500

-4250

0

03570

105140175210245280315350

DOSEPK SamplePredicted PKCL Partial DerivativeVd Partial Derivative


Clinical Trial SimulationsPK/PD Model

Zero Order Infusion over 30 minutes

CL

Vc0.75

70WeightCL TVCL ⎛ ⎞= ⎜ ⎟⎝ ⎠i

1

70WeightVc TVVc ⎛ ⎞= ⎜ ⎟⎝ ⎠i

Scaling Model

γ

γ γ

Emax Conc100 1EC50 +Conc

PD⎛ ⎞⎛ ⎞⋅

= −⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

i

100 %

3 5 7 9 11 13BODYWEIGHT

CL

0-1.5M1.5-3.5M3.5-6.5M6.5-12.5M12.5-24.5M

Covariate Model


Clinical Trial SimulationsDosing Scenarios

Goal

Find the dosing regimen that has the highest likelihood to keep 95% of the patients concentrations above a critical threshold.

Constraints

Dose should be a multiple of the one available strengthDosing may be every two weeks or less


0.0

0.002

0.004

0.006

0.008

0.010

0 500 1000 1500

Regimen 1

0 500 1000 1500

Regimen 2

0 500 1000 1500

Regimen 3

PK

Den

sity

Success Rates (>95% above threshold) of Different Regimens

65% 90% 99%

0.92 0.94 0.96 0.98 1.00Regimen.3

0

20

40

60

80

100


Case Study: Conclusion

Combining Modeling and Simulation tools:

• Understanding of the underlying disease and adult/pediatric differences

• Allometric Scaling

• Optimal design

• Clinical Trial Simulation incorporating variability and uncertainty

• Testing multiple what if Scenarios of dosing Regimens

Made possible to find an effective dose and an optimal design for a prospected pediatric study.

Early clinical findings from the actual trial are confirming the proposed dosing regimen that was based on a computer-assisted trial design.


The Status of MBDD in the U.S. Japan, Korea and China


Status of MBDD: Key Themes From ACoP 2008

Inaugural ACoP Meeting ● 300+ participants from

>100 global sponsors & universities

● MBDD is maturing as a science

● Recognition of key elements for success: interdisciplinary collaboration, organizational support, emergence of standards

● Need to train more practitioners (methods, tools, communication)

● Strong FDA presence and engagement

The inaugural American Conference on Pharmacometrics (ACoP) Meeting, held in March 208, provided a useful backdrop for examining the evolution of MBDD, as well as current challenges and opportunities.


Impact of Pharmacometric (PM) Reviews at FDA

MBDD informs FDA analysis and sponsor interactions to:

● Establish evidence of safety and efficacy

● Optimize dose● Improve future trial

designs

Of 31 NDAs reviewed in 2005-2006, PM was involved in approval for 20 NDAs and labeling for 17 NDAs

● Six published case studies

These data build on earlier FDA published work

● 42 NDAs surveyed between 2000-2004

Impact of Pharmacometric Reviews on New Drug Approval and Labeling Decisions – a Survey of 31 New Drug Applications Submitted Between 2005 and 2006. Bhattaram VA et al. Clin Pharmacol Ther; Volume 81, Number 2. February 2007.


MBDD Metrics Are Becoming the Norm by Which Pharma and Biotechs, Large and Small, Compare and Evaluate Performance

Conclusion: top firms are using model-based methods on 10-20% of projects


Informal Adoption Metrics for Model-based Drug DevelopmentA search of Google Scholar for Nonmem + USA, Nonmem + Japan, etc., shows the following counts of publications from all types of authors (commercial and academic):

0

500

1000

1500

2000

2500

3000

Nonmem +USA

Nonmem +Japan

Nonmem +China

Nonmem +Korea

Goo

gle

Scho

lar

Refe

renc

es

Conclusion: together all kinds of authors (commercial and academic) from Japan, China, and Korea have about the same number of NONMEM-referencing publications as do US authors.


Informal Adoption Metrics for Model-based Drug DevelopmentA search of Google Scholar for Nonmem + Pfizer, Nonmem + Eisai, etc., together with phase I compounds reported in BioPharm Insight shows that ● Top Japanese firms have published about 10% as often as top western firms● The ratio of citations to phase I compounds is 7:1 in the West, and 1:1 in Japan

Conclusion 1: Japanese commercial firms have only about 10-15% of the experience of Western firms in the use of model-based drug development.

Conclusion 2: Since Western firms are using model-based methods in 10-20% of projects, then Japanese companies are using model-based methods in 1-3% of projects.


1st Japanese Conference on Pharmacometrics (JCoP)

Date: November 15 and 16, 2008

Location: Showa University

The President of the Conference: Dr. Syozi Awazu, former President and Professor of Tokyo University of Pharmacy

Director of Secretariat: Prof. H. Sato of University of Showa


Increased Outsourcing Fueling Growth in Asia

Increased Outsourcing to CRO’s and

CMO’s. Overseas CRO’s entering the US market in increasing numbers. Outsourcing includes R&D outsourcing to countries with strong international IP protection and lower cost of business. Latin America and Asia are important because of increasing customer interest outside the US, especially in developing countries. For CRO's, presence and infrastructure in developing regions can be a ‘make or break’ for winning large, global opportunities. Latin America and Asia offer large patient pools with fewer competing studies in most indications. Challenges in these areas include health insurance coverage and hidden trial costs, language and cultural barriers, IP protection, rapid growth and sheer volume, larger and more complex studies and regulatory is sometimes unpredictable.




China is Gaining Ground

Fundamental Trends:

Globalization of Biotechnology: China

has overtaken India in an important

measure of drug research activity. In an

analysis done by the Financial Times, China

now has 274 registered clinical trials

underway, compared to 260 in India. The

trend reflects intensifying interest by the

healthcare sector in China, which is growing

rapidly as a result of rising income and

expanding health coverage and is already

forecast to be the world's fifth-largest

pharmaceuticals market by 2010.



Based on a recent visit to China, I learned a major

(~$1M) investment is planned for

Chinese FDA, with possibility of

creation of a Pharmacometrics

group


Food and Drug Administration Opening Offices in China

Last week, Health and Human Services Secretary Michael Levitt announced that finally, after months of negotiation and bureaucratic slogging, the FDA will be sending permanent staff to China. The first office will be in Beijing and two more offices, one in Shanghai and the other in Guangzhou, will open sometime next year, for a total of eight FDA staffers.

WashingtonPost.com

By Annys Shin | October 20, 2008; 12:11 PM ET


The Future of MBDD in Drug Development


Drivers for Increased Use of MBDD

•Safety issues will drive more individualization of dosing

•There will be increased efforts to identify patients likely to respond to a drug via utilization of genomic or other information

•Companies will have an increasingly greater focus on biotech drugs (large molecules). These are often more complex to develop.

•To become more efficient and develop better targeted drugs, there will be increased emphasis on Translational Medicine / Science. Universities, Publications, Pharmamaceutical companies are embracing this.

● GSK and other companies are already organizing along these lines● Increased emphasis on effective utilization to support MBDD● There will be increased outsourcing and strategic alliances

GlaxoSmithKline, M. D. Anderson Formalize Collaboration in Cancer ResearchNew Alliance to Enhance Integration of Basic Science, Clinical Research

M. D. Anderson News Release 11/05/07


A Vision for MBDD at FDA

Presentation by Dr. Joga Gobburu, FDA, ACoP 2008.


A Path Forward – Companies Need to Make Investments in …

•Increased emphasis and utilization of Biomarkers, including imaging and other techniques

•Implementation of tools to facilitate Translational Science and individualization of patients

•Increased understanding and communication of benefit/risk

•Modeling and Simulation, as a first step towards implementation of MBDD

•Technology, including data repositories and M&S tools

•Shift to Biotechnology

•Academia, to provide programs for training of M&S scientists

•Ongoing training for current staff, to improve their skills

Business

Staff


Acknowledgements

•Joga Gobburu, Pravin Jadhav and Bob Powell, FDA

•Mark Hovde, Adam Rutkin and JF Marier, Pharsight


Thank You

paving the critical path final japanjcop/h20/download_first_congress...supporting "free market...

Documents