biomarker sops - cambridge innovation institute · biomarker sops: getting optimum value from your...

C A M B R I D G E H E A LT H T E C H I N S T I T U T E

ADVANCES REPORTST H E S O U R C E F O R B I O M E D I C A L I N T E L L I G E N C E

CH

CHB

IOM

AR

KE

RSO

PS

JAN

UA

RY

2007

Biomarker SOPs:

Getting Optimum Value from

Your Biomarker Programs

Author: Ken Rubenstein, PhD

Executive Summary

Biomarkers for use in drug discovery and development have grownmarkedly in importance during the past decade, due in large measure to the following factors:

• Pharma’s perceived need to reduce costs and increase NCE(New Chemical Entity) introductions by achieving earlyattrition of unsuitable compounds

• The availability of Genome Era technologies that acceleratediscovery of new biomarkers and facilitate the use ofmulticomponent pattern biomarkers

• Improvements in imaging technologies that facilitate in vivostudies of disease processes and drug targeting

• The rise of the translational medicine paradigm, which usesbiomarkers to help ensure continuity in moving from animalmodels to human subjects

• The increased proportion of programs dealing with new andrelatively unprecedented drug targets

As biomarkers have grown in importance, pharma R&D people havemoved from relatively informal, intuitive ways of dealing withbiomarker discovery and implementation to more formal means, oftenemploying strategic planning, tactical planning, and business analysisfairly early in the discovery stage. The extent to which planning andbusiness analysis are formalized differs among companies, and theorganizational structures used in these processes vary as well. In allpharmaceutical companies, systems for planning, implementation, andemployment of biomarkers remain a work in progress, continuallyevolving as still-scarce information on outcomes of biomarker-drivenprograms collects.

• www.advancesreports.com • Reproduction prohibited vCH

Executive Summary

This report deals with 4 primary issues:

1. Strategic planning for biomarkers in drug discovery anddevelopment

2. Tactical planning for the implementation of biomarkers in programs

3. Organizational structures for biomarker implementation4. Approaches to risk/cost-versus-benefit analysis for biomarker

programs

In gathering information to address these points, we have relied heavilyon semiformal interviews with pharma managers and otherknowledgeable individuals and on a survey conducted among a moregeneral population of pharmaceutical and biotechnology R&Dpersonnel.

The current paradigm in pharma specifies that biomarkers are not beingused merely to assuage scientific curiosity but to answer specificquestions that provide decision-support data as a compound or seriesmoves down the pipeline to increasingly expensive stages ofdevelopment. Such questions generally correspond to a series ofbiomarker categories. Companies differ in their biomarker taxonomies,but the following is a common breakdown into 4 categories (Hurko O. Drug Disc World. 2006;7:63–74).

1. Clinical response biomarkers that identify subjects who will beoptimally responsive to a drug in initial human testing. This isespecially important when a drug addresses a high-risk targetthat has never been proved in humans.

2. Mechanism-of-action biomarkers that directly quantify drug-target interactions. These are especially useful in earlydevelopment to guide initial dosing regimens before the drug-response relationship with respect to clinical outcomes hasbeen established.

3. Efficacy biomarkers that demonstrate the relevance of a drugcandidate to the pathophysiology of the disease in question. Inthis sense, the biomarker serves as a surrogate for the clinicalendpoint, even though it may not have achieved official statusas a surrogate endpoint. Such biomarkers can provide supportfor advancing or dropping a drug, especially one that addressesa high-risk target.

vi • www.advancesreports.com • Reproduction prohibitedCH

Biomarker SOPs: Getting Optimum Value from Your Biomarker Programs

4. Safety biomarkers that can exclude subjects who might reactadversely to a drug or at least identify an adverse reaction atan early stage in a subject receiving a drug. Another type ofsafety biomarker characterizes toxicity and can be used inpreclinical studies to predict whether a compound will betoxic in humans.

Biomarkers can also be divided into those that relate to the compoundor drug mechanism in question and those that relate strictly to thedisease process. The latter often act as surrogates for a clinicalmeasurement and allow monitoring of the response to therapy withoutwaiting for, for example, death to intervene. Biomarkers in thiscategory become candidates for precommercial development, andseveral consortia have formed to discover and develop methods for such biomarkers. The latest and most significant of these is theFDA/NIH/PhRMA (Food and Drug Administration/National Institutesof Health/Pharmaceutical Research and Manufacturers of America)Biomarkers Consortium, which was officially launched October 5,2006. As with the SNP (Single Nucleotide Polymorphism)Consortium, participating pharmaceutical companies will contributefunds so that they, possibly together with academicians and diagnosticscompanies, can identify and develop assays for biomarkers to be used byany or all participants. These biomarkers will be mainly or entirelydisease related.

A second major external factor is affecting both the way biomarkers areused and the nature of the drug discovery and development paradigm.Although it has always been possible in the United States to apply forINDs (Investigational New Drug Applications) that permitmicrodosing of humans with drug candidates based on minimalsupportive data, the FDA’s CDER (Center for Drug Evaluation andResearch) unit published a guidance in January 2006 clarifying whathave come to be called exploratory INDs, and encourages their use. The guidance clarifies microdosing limited numbers of subjects to helpidentify pharmacokinetic and mechanistic properties of compoundsbefore starting formal Phase I trials. The guidance was given highpriority since it is viewed as supporting the FDA’s Critical PathInitiative. Pfizer and several other pharma companies are alreadyconducting such “Phase 0” trials, and many more are in the planningstage. The exploratory IND concept may also permit biomarker-enabledstudies of efficacy for unprecedented targets with limited numbers ofpatients enrolled in Phase I studies.

• www.advancesreports.com • Reproduction prohibited viiCH

Executive Summary

The increasing use of novel, unprecedented targets, about whichrelatively little is known, challenges the classical drug discovery anddevelopment pipeline sequence of activities. More than half of allprograms now address such targets, which carry the burden of increasedrisk of failure in late-stage clinical trials. According to Orest Hurko,MD, Wyeth’s Assistant Vice President, Translational Research, and amember of the Advances Reports Advisory Board, major reasons forsuch failure include the following:

• A drug is administered to the wrong subjects, that is, amixture of responders and nonresponders.

• The drug is given at the wrong dose: Either too little isadministered for adequate receptor occupancy or too much isgiven so that receptors are saturated and excess drug is morelikely to cause off-target effects.

• Indirect efficacy indications provided by clinicalmeasurements are either too noisy or come too late foradequate initial testing of new and unproven targets.

• Some patients become ill from drug exposure.

Strategic planning for biomarkers attempts to answer these questionsfor a given program. Planners must decide such issues as whether aclinical measurement will suffice, whether an existing validatedbiomarker can do the job, or whether a new biomarker must bediscovered and a method for its measurement developed and validated.If a biomarker is indicated, it will usually be intended to provide thefollowing information:

• Biomarkers for identification of optimally responsive subjectsfor initial clinical testing. These are particularly beneficial forhigh-risk targets that have never been proved in humans.

• Biomarkers that permit direct quantification of drug-targetinteractions for selection of initial dosing regimens, especiallyin early-stage development before the drug-responserelationship for clinical effects has been established.

• Efficacy biomarkers that can cost-effectively demonstrate therelevance of an unproven drug target to the pathophysiologyof the disease under consideration, before committing majorfinancial resources to large registration studies with clinicalendpoints.

viii • www.advancesreports.com • Reproduction prohibitedCH


• Safety biomarkers that ideally can be used to exclude subjectswho are likely to become ill after exposure to the drug inquestion, or at least that can identify adverse reactions earlyon after dosing so that subjects can be withdrawn from thestudy before becoming ill.

Although big pharma companies differ somewhat in the methods andtiming for biomarker strategic planning, they generally either start suchplanning very early in the discovery process or intend to do so in thenear future. In some instances, planning and implementation forefficacy-related biomarkers actually begin before a final disease target isdetermined. Since getting a new biomarker on-line can take a year ormore of effort, planning for biomarkers to be used in early developmentclearly needs to begin as soon as strategic issues can be clarified.Planners must also decide early on whether to develop biomarkers in-house or to outsource the effort.

The types of biomarkers and levels of validation differ depending onthe stage of a program. Essentially all scientific experiments done inpharmaceutical R&D require a molecular or functional readout thatcould formally be considered a biomarker. Many such experiments areconducted at the prediscovery stage in the process of characterizingpotential targets. The readouts from such experiments are not usuallyconsidered biomarkers. However, as mentioned earlier, disease-relatedbiomarker development may actually precede target identification.During the discovery phase, mechanism-of-action biomarkers areparticularly useful to verify that a compound is hitting its intendedtarget, whether using cells in culture, tissues, or intact animals. Thedegree of validation of a biomarker typically increases in stringency as a compound or series advances down the pipeline.

Biomarkers become especially valuable at the preclinical developmentstage of a program, when spending rates start to accelerate markedly.Preclinical studies with animal models often use biomarkers as a meansto strengthen predictability of successful translation to humans. In fact,translational medicine departments in pharma are increasinglybecoming centers of biomarker planning and development. Biomarkersfor preclinical development are typically related to pharmacodynamics,efficacy, and safety. Naturally, biomarkers that are suitable for use inanimal models often cannot be translated to human clinical studies,since specimens may be collected from animal sites that are notaccessible in humans (e.g., brain tissue). Therefore, although it may bepermissible to do gene expression profiling on tissue collected fromanimal models, this must often be replaced with protein or metabolite

• www.advancesreports.com • Reproduction prohibited ixCH

Executive Summary

signatures accessible in biological samples from humans. Biomarkersbased on in vivo imaging studies can be particularly useful intranslation if they are equally applicable to animals and humans.

Much of the biomarker work in human studies occurs in Phase 0 orPhase I development. Biomarkers to aid in dose-ranging and to confirmthe mechanism of action have their greatest usefulness in these earlystages, whereas biomarkers to select patients who are most likely torespond to a drug and to predict adverse reactions become useful inPhases II and III. Usually, the proof of concept for a drug candidatemust await Phase II studies, but increased use of high-qualitybiomarkers to predict efficacy suggests that, in some cases, patients canbe included in Phase I trials. Clinical endpoints for efficacy monitoringare preferred when possible, especially since validating biomarkers forstatus as surrogate endpoints can be a long and arduous process.

Pharmaceutical companies employ one of several organizationalstructures to deal with biomarkers. Wyeth’s Dr. Orest Hurko describes 3 major models:

1. The Explicit Model, where the biomarker or translationalmedicine group stands independently alongside the discoveryand clinical development organizations

2. The Implicit Model, where biomarker activities are absorbedinto pre-existing research and development organizationalentities

3. The Hybrid Model, where biomarker activities are segmentedindependently, but report in a matrix arrangement to discoveryand clinical development organizations

Bristol-Myers Squibb employs something approximating the ExplicitModel, in which biomarker work centering on the translation frompreclinical development through proof of concept in early developmentis centralized. However, early discovery and safety biomarker work aredecentralized into functional organizations. AstraZeneca employs an Implicit Model, in which most biomarker planning andimplementation work are done within existing functions, although a separate group is charged with bringing new biomarker-relatedtechnologies into the company. Lilly employs a Hybrid Model in whicha governing Biomarker Working Group assures that project teamsproperly formulate a biomarker strategy and plan for its timelyimplementation. Pfizer also employs a Hybrid Model, in which atranslational medicine group deploys representatives into therapeutic

x • www.advancesreports.com • Reproduction prohibitedCH


area teams to influence the course of biomarker needs. Thetranslational medicine department retains budgetary control forbiomarker outsourcing.

Once the biomarker strategic plan is in place, researchers need toassemble a tactical plan for implementation of biomarkers in theirproject. The tactical plan must include the means for biomarkeracquisition (if possible), biomarker discovery (if necessary), methoddevelopment, and biomarker validation. Whereas strategic plans varyconsiderably according to the needs of the project, tactical plansbecome increasingly formatted and subject to company guidances and standard operating procedures.

Since validation of biomarker assays is critically important, it hasbecome the subject of intense consideration, and this focus has led to the emergence of the fit-for-purpose validation concept. Anintercompany group recently published a position paper setting forthguidelines for 4 levels of validation, in order of increasing rigor: (1) pre-analytical and method development, (2) exploratory methodvalidation, (3) advanced method validation, and (4) in-studyvalidation (Lee JW et al. Pharm Res. 2006;23:312–328). Assayparameters considered at each validation level include dynamic range,sensitivity, selectivity and specificity, analytical precision and accuracy,and biological recovery/accuracy. Details for each level are presented in Chapter 3.

Companies differ in the degree to which they have centralized andformalized method development and validation. For example, Bristol-Myers Squibb has a Biomarker Best Practices guidance that allows agreat deal of latitude depending on the needs of the project. Eli Lillyprovides its researchers with a brief guideline setting forth mainlyobjectives and expectations for biomarkers in a program. Pfizer providesan extensive guideline to its project teams that includes details on allmanner of issues and perspectives that may need to be considered indeploying biomarkers for a project. A description and extracts from this document can be found in Chapter 6.

Companies also differ in their decision processes for developingbiomarkers in-house versus outsourcing. For example, although Bristol-Myers Squibb generates biomarkers for early development primarily in-house, it does do some outsourcing. Decisions are based primarily oncost-effectiveness. Lilly develops essentially all biomarkers for use upthrough the preclinical phase in-house; biomarkers for later stages may

• www.advancesreports.com • Reproduction prohibited xiCH

Executive Summary

be developed either way. However, Lilly has chosen not to establish thecapability to develop imaging biomarkers in-house. Pfizer does a limitedamount of imaging biomarker development in-house, choosing tooutsource it for the most part. For molecular biomarkers, the decision is made for each case, considering in-house resources and capabilities,cost, quality if outsourced, track record of potential vendors, need tomaintain control, and speed. Pfizer and others now add the option ofgiving development of some biomarkers over to the emergingprecommercial Biomarkers Consortium.

In viewing organizational structures for biomarker work, we considerthe 3 models delineated by Wyeth’s Orest Hurko (see above). As wasmentioned, Bristol-Myers Squibb exemplifies the Explicit Model,AstraZeneca the Implicit Model, and Lilly and Pfizer the HybridModel. Details for each organization are presented in Chapter 4.

When researchers decide they need to discover new biomarkers ordevelop new assays, they may require funding that can run into themillions of dollars spent over a year or more. Clearly, when time andcost start to mount, some level of business analysis can be beneficial.This can take the form of ROI (return on investment) analysis,cost/risk-versus-benefit analysis, or both. Pfizer is one of the few pharmacompanies to provide its project teams with a formal software tool forROI analysis and to encourage its use. Although gaining momentum, ithas not yet been universally adopted by project teams. As demonstratedin the Pfizer case study in Chapter 6, business analysis in biomarkerprograms can aid in choosing whether to take on a biomarker program,whether to do the work in-house or outsource it, or even whether towait for an industry consortium to provide a solution. Unfortunately,many programs employing new-wave biomarkers are still indevelopment, so that ultimate benefits (or lack thereof) of biomarkersare not yet visible for use in business analysis.

xii • www.advancesreports.com • Reproduction prohibitedCH

Table of Contents

• www.advancesreports.com • Reproduction prohibited xiii

CHAPTER 1

INTRODUCTION ......................................................................................1

1.1. Historical Perspective on Biomarkers in Pharma ..............................3Terminology Recently Evolved ............................................................4A Shift in Emphasis ..............................................................................5

1.2. Current Status of Biomarkers in Pharma ..........................................7

1.3. Planning and Implementation Issues ................................................10Strategic Planning Issues ....................................................................10Tactical Planning Issues ......................................................................11Organizational Considerations ..........................................................11Business Considerations......................................................................12

1.4. Precommercial, Collaborative Biomarker Work ..............................12The Biomarkers Consortium ..............................................................12

Consortium Projects ....................................................................15European Union Consortium ............................................................16

1.5. Exploratory Investigational New Drug Applications andthe Pipeline Paradigm ......................................................................18

CHAPTER 2

BIOMARKER STRATEGIC PLANNING ............................................23Biomarkers Can Serve Multiple Purposes ..........................................25

2.1. Planning for Biomarkers ....................................................................26Variations among Therapeutic Areas ................................................28

2.2. Biomarkers for Stages along the Pipeline ........................................28Prediscovery ........................................................................................28Discovery ............................................................................................30Preclinical............................................................................................32Early Clinical Development ..............................................................35Late-Stage Clinical Development ......................................................36Efficacy Biomarkers ............................................................................37

2.3. Postmarketing......................................................................................39

2.4. The Biomarker Strategic Plan ..........................................................39

CH

Table of Contents

CHAPTER 3

BIOMARKER IMPLEMENTATION PLANNING..............................41

3.1. Biomarker Validation..........................................................................41Pre-analytical Considerations ............................................................42Method Development ........................................................................43Exploratory Method Validation..........................................................45Advanced Method Validation ............................................................46In-Study Validation ............................................................................46

3.2. Industry Experts’ Comments on Biomarker Guidelines,Standard Operating Procedures, and Validation..............................47

3.3. In-House versus Outsourced Biomarker Identification andMethod Development ........................................................................50The Biomarkers Consortium ..............................................................52Example of Biomarker Implementation ............................................54

CHAPTER 4

ORGANIZATIONAL ISSUES ................................................................57Translational Medicine ......................................................................58Globalization versus Decentralization ................................................62

4.1. Explicit Model ....................................................................................63

4.2. Implicit Model ....................................................................................64

4.3. Hybrid Model ......................................................................................67

CHAPTER 5

THE BIOMARKER BUSINESS CASE..................................................73

5.1. Comments on Business Analysis from Industry Experts ................74Cost for One, Benefit for Another ....................................................77

CHAPTER 6

PFIZER: CASE STUDY IN BIOMARKER PLANNINGAND IMPLEMENTATION ............................................................79

6.1. Biomarker Typology and Linkage to Outcome: Target,Mechanism, and Outcome Biomarkers ............................................79

6.2. Validation Typology ............................................................................81

6.3. Stages in the Biomarker Life Cycle: Pathfinding,Research, Development ....................................................................82

6.4. Business Considerations: Expense of Developmentversus Cost of Wrong Decision ........................................................83

xiv • www.advancesreports.com • Reproduction prohibitedCH


6.5. Biomarker Best Practices: Optimize, Maximize, and Balance ........85

6.6. Biomarker Validation..........................................................................85

6.7. Minimally Acceptable Criteria ..........................................................89

CHAPTER 7

RESULTS OF BIOMARKERS SURVEY ..............................................93

CHAPTER 8

OBSERVATIONS AND CONCLUSIONS ..........................................105

APPENDIX

EXPERT INTERVIEWS ........................................................................107Ernie Bush, PhD, Director New Initiatives, CambridgeHealthtech Associates ..........................................................................107Claudio Carini, MD, PhD, Vice President, Translational Medicine, Development & Regulatory Services, MDS Pharma Services ..............117Cynthia S. Cheesman, Assistant Vice President, PreclinicalProject Management, Wyeth ..............................................................122Nicholas Dracopoli, PhD, Vice President, Clinical DiscoveryTechnologies, Bristol-Myers Squibb......................................................125Darrick Fu, MBA, Associate Vice President for Science andRegulatory Affairs, PhRMA ................................................................128Orest Hurko, MD, Assistant Vice President, TranslationalResearch, Wyeth ................................................................................131David S. Lester, PhD, New York Site Head, Pfizer Human HealthTechnologies, Global Clinical Technology—Pfizer Global Researchand Development ................................................................................135Terry Lindstrom, PhD, Distinguished Research Fellow;Drug Disposition, Global Pharmacokinetics, and Toxicology,Eli Lilly & Co. ..................................................................................136Bruce H. Littman, MD, Global Head of Translational Medicine,Pfizer ..................................................................................................143Michael Stocum, MS, Managing Director, PersonalizedMedicine Partners ..............................................................................150Stephen A. Williams, MD, PhD, Vice President and WorldwideHead of Clinical Technology, Pfizer ....................................................156

References ..................................................................................................163

Company Index with Web Addresses ..........................................165

• www.advancesreports.com • Reproduction prohibited xvCH

Table of Contents

FIGURESFigure 6.1. Evolution of a New Biomarker ......................................................84

Figure 7.1. Survey: Respondents by Sector ......................................................93

Figure 7.2. Survey: Respondents by Position....................................................94

Figure 7.3. Survey: Respondents by Stage of Work ........................................95

Figure 7.4. Survey: Respondents by Research Focus ......................................96

Figure 7.5. Survey: Involvement with Biomarker Planning and Discovery ....97

Figure 7.6. Survey: View toward Precommercial Biomarker Development ....98

Figure 7.7. Survey: View toward Standard Operating Procedures orGuidelines for Biomarker Development ..........................................................99

Figure 7.8. Survey: View toward Outsourcing Biomarker Discoveryand Development ..............................................................................................99

Figure 7.9. Survey: Reasons for Biomarker Outsourcing ................................100

Figure 7.10. Survey: Types of Biomarkers Used ..............................................101

Figure 7.11. Survey: Stage at Which Biomarkers Are Used ..........................102

Figure 7.12. Survey: Fiscal 2007 Biomarker-Related Budgetary Plans ..........103

Figure 7.13. Survey: Three-Year Biomarker-Related Budgetary Plans ..........104

TABLESTable 1.1. GlaxoSmithKline’s Biomarker Typology............................................9

Table 6.1. Purpose versus Characteristics of Diagnostics Used asScreening Tools ................................................................................................86

Table 6.2. Biomarker Validation Attributes ....................................................88

xvi • www.advancesreports.com • Reproduction prohibitedCH

Chapter 1INTRODUCTION

Biomarkers have come to play several important roles in drug discoveryand development, and their importance is still increasing. This trend isdriven by 2 main factors: (1) Omic technologies developed during thepast decade provided powerful new means for discovery of newbiomarkers, and (2) more recently, the development and disseminationof convenient multiplex assay technologies have facilitated the use ofmulti-analyte biomarkers derived from omic studies, thereby addingconsiderable diversity to the analytical possibilities available forbiomarker programs.

From the business perspective, the well-documented discrepancybetween pharma R&D expenditures and new product approvals hasdriven the search for ways to increase early attrition of unsuitable drugcandidates and reduce expensive failures in late-stage clinical trials. In addition, pharmaceutical companies have been forced by economicrealities in recent years to shift emphasis away from me-too drugprograms toward reliance on the plethora of new or relativelyunprecedented drug targets arising in the postgenomic period. Thepaucity of biological knowledge available for most of these targetsstresses the importance of biomarkers to guide drug discovery anddevelopment scientists in their decision-making.

Pfizer’s Dr. Bruce Littman (see interview, Appendix), when asked aboutnew unprecedented targets, stated: “Clearly, they make up the majorityof the portfolio in most large pharma companies now. You can’t developa me-too drug anymore and expect to be able to sell it at a good price,and one of the best ways of trying to get differentiation is to use new,unprecedented drug targets.”

• www.advancesreports.com • Reproduction prohibited 1CH

Introduction

Early efforts to increase the use of biomarkers in drug discovery anddevelopment tended to be ad hoc in nature and often did not benefitmuch from detailed strategic planning, tactical planning, or cost/risk-versus-benefit analysis. As the need for and usage of biomarkers havebecome better established in pharma, so have the levels of planningand business justification afforded them. However, such efforts are stillvery much a work in progress. Systems for planning andimplementation differ considerably among companies, as do levels ofsophistication in using biomarkers and the degrees of planning andbusiness analysis applied.

This report deals with 4 primary issues:

1. Strategic planning for biomarkers in drug discovery anddevelopment

2. Tactical planning for the implementation of biomarkers inprograms

3. Organizational structures for biomarker implementation4. Approaches to risk/cost-versus-benefit analysis for biomarker

programs

In pursuit of investigating these issues, we interviewed a number ofmanagers and staff people from large pharmaceutical and biotechnologycompanies. Transcripts from 11 of these interviews are included in theAppendix. We also conducted an on-line survey of knowledgeableindustry participants in order to shed further light on current planningand organizational practices with respect to biomarkers. Finally, wedrew background information from secondary sources relevant to oursubject. These include journals, trade magazines, and conferenceproceedings.

This report is organized into 9 chapters. Chapter 1 delineates thepurpose and organization of the report along with an historicalperspective on the role of biomarkers in pharma, background materialon why and how biomarkers are used in pharma today, and furthercommentary on the 4 key issues to be examined.

Chapter 2 addresses the first of these 4 issues: strategic planning, withemphasis on identifying the biomarker requirements for drug discoveryand development programs. Chapter 3 addresses tactical planning forimplementation of biomarker strategies. Chapter 4 deals with the

2 • www.advancesreports.com • Reproduction prohibitedCH


organizational structures that companies deploy in dealing withbiomarker planning, project control, and implementation. Strategicand tactical planning for biomarker utilization in the context of anorganizational structure pave the way for business analysis of the costsand risks versus benefits of proposed biomarker work, which is thesubject of Chapter 5.

Chapter 6 provides a case study comprising an in-depth look at Pfizer’sguidelines and philosophies with respect to biomarkers for drugdiscovery and development. Chapter 7 presents data and conclusionsfrom a market survey of pharmaceutical and biotechnology industrypeople active in biomarker work. Chapter 8 focuses on conclusions andobservations from information given in this report, and the Appendixprovides transcripts of interviews with industry experts, portions ofwhich appear in context earlier in the report.

1.1. Historical Perspective on Biomarkers in Pharma

Biomarkers have played an important role in drug discovery anddevelopment for nearly as long as the pharmaceutical industry has beenin existence. To appreciate the meaning of this statement, one canreview the currently accepted definitions of a biomarker and associatedentities. The following 3 formal definitions come from the BiomarkersDefinition Working Group of the US National Institutes of Health(NIH), which published the results of its deliberations in 1999 onbiological measurements in therapeutic development.1

• Biomarker: “A characteristic that is objectively measured and evaluated as an indicator of normal biological processes,pathogenic processes, or pharmacologic response to atherapeutic intervention.”

• Clinical endpoint: “A characteristic that defines how a patientfeels, functions, or survives.”

• Surrogate endpoint: “A biomarker intended to substitute for a clinical endpoint.”

The following is a somewhat less formal and, perhaps, more operationaldefinition: “Any biological measurement that provides actionableinformation regarding disease progression, pharmacology, or safety thatcan be used as a basis for decision-making in drug development.”2

Biomarkers have

played an

important role in

drug discovery

and development

for nearly as

long as the

pharmaceutical

industry has been

in existence.


Introduction

Two related terms deserve definition in our current context:

• Analytical method validation: “The process of assessing theperformance characteristics of a given assay.”2

• Clinical qualification: “The evidentiary and statistical processlinking biologic, pathologic, and clinical endpoints to the drugeffect, or linking a biomarker to biologic and clinicalendpoints.”3

These last 2 definitions are critically important. The first deals withassuring that a biomarker assay measures what it purports to measurewithin defined limits, whereas the second deals with the question ofwhether the biomarker is useful for its intended purpose.

Terminology Recently Evolved

Colorimetric assays to measure enzyme activities and metabolite levelscame available early in the 20th century. Since that time researchersand clinicians have used biomarker assays to determine physiologicalstatus, diagnose diseases, and monitor their progress. For much of thelast century, pharmaceutical scientists have used many of these sameassays to monitor the safety and efficacy of drug candidates. However,the term biomarker rarely, if ever, came into play until recently. Many of the assays were drawn from the standard clinical chemistryarmamentarium, while others were devised strictly for use inpharmaceutical R&D.

Although assays to measure enzyme activity and metabolite levels inbiological fluids at relatively high concentrations were generally simpleand straightforward to develop and validate, assays for proteins andsmall molecules present in submicromolar quantities often posed greaterdifficulty. For example, many of these molecules could only bemeasured by immunoassays, and if suitable antibodies were notavailable, they needed to be generated via elaborate and expensiveprocesses, sometimes requiring a year or more.

Furthermore, in order to discover new biomarkers based on proteinspresent at micromolar to nanomolar levels, researchers were oftenrequired to start with very large quantities of animal tissue and gothrough painstaking multistep procedures to isolate minute quantitiesof material for antibody generation and for use as assay standards.Consequently, only a modest number of such biomarker assays were available by the close of the 20th century.



The 21st century renaissance in biomarker discovery and utilization was driven by 2 major factors: (1) the pharmaceutical industry’sgrowing need to get more drugs to market faster and at lower cost thanpreviously and (2) the availability of omic and other Genome Eratechnologies to facilitate biomarker discovery and implementation.Pharma’s innovation deficit has been sufficiently well documented that we need not deal with it here in any detail.

Regarding the second driver, omic technologies began to emerge in the late 1990s and captured the imagination of both academic andpharmaceutical industry researchers as means for discovering newdisease genes, drug targets, toxicity signatures, mechanism signatures,and the like. Although many new potential drug targets have beenidentified, progress toward bringing them into productive use has beenslower and more awkward than originally anticipated. Whilebiomarkers have not yet, arguably, lived up to their full potential inguiding the successful use of these new unprecedented targets, progressin using biomarkers to enable early compound and target attrition and,generally, to aid decision-making in drug discovery and developmenthas been sufficient to impact organizational structures and R&Dparadigms. Changes include the formation of new functional groupsand matrixed structures dedicated to planning and implementingbiomarker strategies.

A Shift in Emphasis

The introduction of commercial DNA microarrays in the late 1990sstimulated the application of transcriptomics in drug discovery. Theability to view up- and down-regulation of thousands of genessimultaneously in disease versus health, drug versus drug, drug versuscontrol, and so forth, evolved into the notion that groups ofdifferentially regulated genes could be used as signatures for particularbiological events. From this realization, the general notion of abiomarker shifted from emphasis on a single molecule signaling singleevents to multiple molecules signaling single events. Although DNAmicroarrays might not often be the perfect delivery vehicle for theroutine assay of these multiplex gene signatures, other means, such as real-time quantitative polymerase chain reaction (PCR) or beadarrays, were often more appropriate.

Successes (and limitations) of transcriptomics stimulated efforts todevelop proteomics, in which shifts in large numbers of proteins couldbe viewed in single experiments. Although technologies for “proteome-wide” studies proved to be considerably less agile than those fortranscriptomics, the field slowly gained ground and has become

The introduction

of commercial

DNA microarrays

in the late 1990s

stimulated the

application of

transcriptomics in

drug discovery.


Biomarker Implementation Planning

validation level include dynamic range, sensitivity, selectivity andspecificity, analytical precision and accuracy, and biologicalrecovery/accuracy.

Key items for each validation level are presented in the followingsections.

Pre-analytical Considerations

For each biomarker to be used in a program, researchers should firstcreate a biomarker work plan. The level of detail required for the planwill naturally depend on whether they propose to use an establishedbiomarker for which off-the-shelf reagents are available or whether theyneed to develop a new assay for a previously identified biomarker. Anywork needed to discover a new biomarker for use in answering aspecific question in a discovery or development program must becompleted before laying out the biomarker work plan. This report doesnot deal specifically with biomarker discovery, a topic that has beencovered in great detail in numerous reports and scientific publications.

The biomarker work plan as envisioned by the AAPS 2003 BiomarkerWorkshop team starts with defining objectives of the study. Theseobjectives support researchers in identifying their needs in the areas of reagents, controls, and experimental samples. The objectives shouldalso include the intended use of the assay, which permits definition of the level of rigor required for assay validation.

For biomarkers to be used in early development, the plan should alsoinclude information gleaned from preclinical studies and literaturereviews that might provide background information for a givenpopulation. These considerations can aid in establishing the precisionrequirements for the assay. For example, an assay measuring levels of a biomarker for bone resorption may already have been studied inindividuals with arthritis or osteoporosis, and data on interdonor andintradonor variability may already exist. These data allow specificationof the minimally detectable changes in the biomarker.3

Sample collection issues may appear to be trivial, but experiencedbioanalysts know that a number of variables must be controlled in orderto assure that an assay result adequately represents the level ofbiomarker present in the sample and is not altered by extraneousvariables. For example, most samples will have to go through at

For each

biomarker to

be used in

a program,

researchers should

first create

a biomarker

work plan.



least 1 freeze/thaw cycle. This usually requires studying the effect of thefreeze/thaw process on results plus studies of sample stability in thethawed state.

It may also be important to define sample collection and handlingprotocols in instances where these operations might introducevariability into the biomarker assay result. Possible patient-relatedvariables include variations in biomarker levels during the course of aday or other time period, variations due to disease subtype or stage, andbehavioral variables due to such factors as diet or emotional state.

For blood collection it can be important to control for such details asthe type of needle used, duration of the draw, collection tube type, andanticoagulant type and level. These and other factors can affect thestability of the biomarker. Any device that contacts the sample needsto be checked to assure that the biomarker in question will not adsorbto its surface, thereby reducing its level in the specimen. In someinstances, plasma may be preferred over serum, and the conditions forcoagulation need to be controlled accordingly. Hemolysis can happenduring sample collection, and its effects on the assay need to bedetermined.

The same principles apply generally to tissue samples. However, tissuesare inherently heterogeneous and therefore require customized samplingtechniques. Standardized protocols for collecting live tissue, which mayrequire sterile technique, often need to be executed by personneltrained and/or certified in histology or pathology. Tissue homogenatesoften require normalization. For example, the quantity of biomarkermay be normalized to tissue mass or protein content, and conditionsmust be defined accordingly.

Method Development

Although methods for assaying a number of established biomarkers arereadily available, it is often necessary when dealing with newerbiomarkers or variant sample types to develop a new method. Naturally,the extent to which an assay must be developed depends on its use. Forexample, a biomarker assay for use in studying the mechanism of actionof a compound administered to cultured tissue will typically require lessrigorous development and validation than one intended for use in dose-ranging studies in early development.


...tissues are

inherently

heterogeneous and

therefore require

customized

sampling

techniques.

Biomarker Implementation Planning

A first step in method development involves establishing the feasibilityof the proposed method by determining if the reagents are of adequatequality and the performance characteristics of the assay match therequirements of the proposed study. For ligand-binding methods aprecision profile (plot of coefficient of variation vs. log calibratorconcentration) is particularly useful to determine if an assay is able tomeasure the biomarker over a particular concentration range.3 It isadvisable at this stage to do preliminary testing for assay selectivity,linearity, and range of quantitation using the assay reagents, the sample matrix, and the calibrator matrix (if relevant).

The sample matrix often affects the assay reading to a different extentthan the calibrator matrix. This may require cleaning up the sample toremove or dilute the matrix sufficiently to permit adequate assayperformance. The presence of the drug of interest in a sample can affectboth the selectivity and accuracy of a biomarker assay, especially whenthe binding of the drug to its target competes with binding of the drugto an antibody used in an assay. If the biomarker in question is alsofound endogenously in the sample matrix, it may have to be strippedout (using charcoal absorption, for example) before calibration materialcan be added back. In extreme examples, a synthetic calibrator matrixmay have to be devised (e.g., buffer plus 1 or more proteins) to mimicthe natural sample matrix. For multiplex assays, it may not be possibleto calibrate accurately for each analyte, and some bias or nonlinearitymay have be accepted for one or more of them.

Care must also be taken in selection of calibration material, samples for validation, and quality control materials. Pure materials of modestmolecular weight make the most reliable calibrators for quantitative,well-behaved assays. Calibration material of high molecular weight(e.g., >1,000 daltons) or with ill-defined homogeneity often requiresanalytical compromises because of ambiguities in defining assayaccuracy. In these circumstances, accuracy can be assumed when assayresults obtained with various dilutions of the calibrator are consistentwith expectation. When accuracy cannot be assured, assays may haveto be downgraded from quantitative to “quasiquantitative” (i.e., relative rather than absolute quantitation) or even qualitative.

Validation samples are used at the assay validation stage to studyprecision, accuracy, stability, and the like. Quality control samples areused during an actual biomarker study to determine if a particular run isbehaving in accordance with expectations. Often the same samples canbe used for both purposes, with the proviso that they should be as

Pure materials of

modest molecular

weight make the

most reliable

calibrators for

quantitative, well-

behaved assays.


Pfizer: Case Study in Biomarker Planning and Implementation


6.3. Stages in the Biomarker Life Cycle: Pathfinding,Research, Development

The guideline goes on to define stages in the biomarker lifecycle, whichhelp in understating how far a biomarker development program has togo in order to be applicable for its intended purpose, and to describethe kinds of activities that should occur at each stage.

• Pathfinding: The earliest stage of technical development,although not all biomarkers start at this stage. A biomarkerbegins life in this stage if it is invented as a technique but anappropriate purpose is not yet clear and objectives are notdefined. The objective of experimentation in this stage is to evaluate multiple opportunities and learn what use thetechnique would have. At this stage, if an invention has beenmade it must be discussed with the therapeutic area and/orbiotech attorney to review whether to apply for a patent.

• Research: The stage at which objectives can be defined,including a proof of concept and an outline of the biomarkerMAC (minimally acceptable criteria). Biomarkers that areinvented for a specific purpose start their life here. The aim is to reach a basic proof of concept for the biomarker in thespecies of application or terminate the biomarker programefficiently. Uncertainty may be high. The biomarker should bereviewed for FTO (freedom to operate) and a decision taken ifany possible third-party infringements can be identified. Thismay be dealt with by seeking a license or developing a work-around.

• Development: The stage at which proof of concept has beenreached and issues of practicality, cost, and performance arebeing evaluated. At this stage, the MAC document becomesmuch more detailed and the biomarker is prospectivelyevaluated against it. Any IP (Intellectual Property) aspectincluded in the overall risk/benefit analysis and any potentiallicense payments included in the project costs are considered.



• Application: The stage at which the minimally acceptablecriteria have been fulfilled. The biomarker is used for itsintended purpose, and the emphasis is on incrementalimprovements in performance and cost and management ofquality. A biomarker that is in this stage for one purpose maystill be in an earlier stage for a different purpose, with adifferent MAC.

6.4. Business Considerations: Expense of Developmentversus Cost of Wrong Decision

From a business perspective, Pfizer states implicit assumptions that mustbe addressed if a biomarker is to meet the stated goal of reducing thecost and increasing the value of decisions. We note that Pfizer employsan ROI (return on investment) software tool designed specifically foruse in biomarker strategic planning, and all teams are urged to makeuse of it. Factors considered include:

• Any increased uncertainty due to application of a biomarkerdoes not lead to an intolerably increased probability of awrong decision.

• The expense of developing the biomarker does not erase the financial advantage versus existing approaches.

• The biomarker must be available in time for the intended use in the program.

From a qualitative perspective, Pfizer uses a graph representing the cost of a decision versus the risk of a false decision (Figure 6.1).

Expert Interviews

In most cases, a biomarker should satisfy 2 criteria. First, it must beassociated with the biological mechanism, disease, or treatment; inother cases, it must also correlate statistically to the clinical outcome.That’s another important aspect. Even more, I would say that thescientist or physicians using or implementing biomarkers should have astrong rationale for why the biomarker is useful and accurately reflectsor predicts a potential outcome. That is another important concept for biomarkers.

I’ll give you an example: CD4-positive cells in HIV can give you anidea of how the therapy is affecting the patient. They can indicatewhether the disease is present, because the biomarker is linked to thedisease as well as to viral load, and they can also be prognostic. So thatis the best biomarker for assessing recent HIV infection, and there aremany more examples. A much more common example is the use ofcholesterol and blood pressure as very important biomarkers forpredicting cardiovascular disease; we’ve been using them for years.

In summary, CROs like MDS Pharma, with large and diversetechnology platforms, can provide an appropriate alternative for thevast majority of companies developing drugs since they can offer a large platform in the area of biomarker strategy.

Cynthia S. Cheesman

Assistant Vice President, Preclinical Project Management,Wyeth

Cambridge Healthtech Institute (CHI): Please start by describing yourposition in the context of biomarker work at Wyeth.

Cynthia S. Cheesman (CSC): Preclinical Project Management atWyeth is a group of project directors who work with the discoveryteams. As discovery research programs on given targets become morerefined, my function is to work with them to facilitate the discovery-development transition phase. My group organizes an early projectteam, which characterizes early synthetic feasibility, physical chemistrycharacteristics, early pharmacokinetics, drug metabolism, and safetyissues, and assures that there’s a clinical development path and amarket for this novel compound. We actually conduct the initial tasksand activities needed to be sure that the discovery moiety will hit theground running, so to speak.



In the last 18 months, we’ve included both a traditional pharmacologyand a translational medicine representative in the project teams earlyin their development phase. These are individuals who may have abackground in genomics, biochemistry, or medicine, whose primaryfunction is to bring to the discovery team the dimension oftranslatability. In some cases, depending on the target, biomarkersmight be readily available to be used as indicators of activity, whereasin others—and this is the harder one, of course—there aren’t anytranslative biomarkers. For many neuroscience targets, for instance,there may not be readily identified biomarkers that we can access forearly activity in our animal models that would carry through todevelopment. So, from a structural point of view, my group assures thata translational medicine resource has been added to our teams and isworking as integrally as possible with the other disciplines withindiscovery.

CHI: Can you elaborate on the nature of the project flow from earlydiscovery through market?

CSC: A key event in our development process is the DevelopmentTrack milestone. Project “ownership” effectively changes hands at thattime, from discovery to development. To limit the possibility for errorand time loss, our organization has established an overlap of discoveryand development processes. Our project team structure is the vehicle bywhich continuity is maintained. A translational research representativefrom within discovery supports programs before placement ondevelopment track, and, shortly before the development trackmilestone, a translational development staff member joins our teamfrom within clinical research. Each team has a translational researchand a translational development representative. I believe that thesystem will be successful, but I also believe some of its success will relyon individual personality. There could be vulnerability in such astructure, because it has 2 separate organizational lines. Success relies a lot on the strength of the teams and mutual respect; those are difficult things to assure.

As you may have heard, in the last calendar year Wyeth has switchedits entire way of running project teams to a paradigm called “Learn andConfirm,” in which a given compound is no longer in a team by itself.We group our compounds by disease categories. A lead leaves discoveryas a single compound, but as it proceeds into development, it is placedinto a team that is looking at multiple leads for the same disease.There’s therapeutic gain, because some translational research and


biomarker sops - cambridge innovation institute · biomarker sops: getting optimum value from your...

Documents