Imaging biomarkers as surrogate endpointsfor drug developmentWolf S. Richter1

1 Schering AG, Global Medical Development Diagnostics, Müllerstraβe 178, 13342 Berlin, Germany

Published online: 24 May 2006© Springer-Verlag 2006

Abstract. The employment of biomarkers (includingimaging biomarkers, especially PET) in drug developmenthas gained increasing attention during recent years. This hasbeen partly stimulated by the hope that the integration ofbiomarkers into drug development programmes may be ameans to increase the efficiency and effectiveness of thedrug development process by early identification ofpromising drug candidates—thereby counteracting therising costs of drug development. More importantly,however, the interest in biomarkers for drug developmentis the logical consequence of recent advances in biosciencesandmedicine which are leading to target-specific treatmentsin the framework of “personalised medicine”. A consider-able proportion of target-specific drugs will show effects insubgroups of patients only. Biomarkers are a means toidentify potential responders, or patient subgroups at riskfor specific side-effects. Biomarkers are used in early drugdevelopment in the context of translational medicine to gaininformation about the drug’s potential in different patientgroups and disease states. The information obtained at thisstage is mainly important for designing subsequent clinicaltrials and to identify promising drug candidates. Biomar-kers in later phases of clinical development may—ifproperly validated—serve as surrogate endpoints for clin-ical outcomes. Regulatory agencies in the EU and the USAhave facilitated the use of biomarkers early in the devel-opment process. The validation of biomarkers as surrogateendpoints is part of FDA’s “critical path initiative”.

Keywords: Biomarker – Surrogate endpoint – PET –Drug development

Eur J Nucl Med Mol Imaging (2006) 33:S6–S10DOI 10.1007/s00259-006-0129-z

The background

During the past two decades, considerable progress hasbeen made in biosciences and medicine, as exemplified bythe sequencing of the human genome [1], the increase in thenumber of genome-based potential drug targets and thesuccessful introduction of target-specific therapeutics likeimatinib (Gleevec) and trastuzumab (Herceptin) into clin-ical routine. In diagnostic imaging, a similar progress hasoccurred, with the broad availability of positron emissiontomography (PET) together with its most important radio-labelled tracer, [18F]-fluorodeoxyglucose (FDG), and therecent introduction of new imaging devices that combinefunctional imaging with PET and excellent anatomicaldelineation with multi-detector computed tomography(PET/CT).

The drug development process

While the progress in biosciences and preclinical research isobvious and numerous potential drug targets have beenidentified, this progress is not (yet) paralleled by a com-parable increase in marketing approvals of new diagnosticand therapeutic compounds. On the contrary, the number ofnew drug applications (NDAs) has dropped during recentyears in the USA as well as in the European Union [2, 3].Research and analyses have been conducted to assess thereasons for the current decline in NDAs and correctivemeasures have been proposed (see, e.g. [3]). It is in thiscontext that the potential value of biomarkers has gainedinterest as a means to increase the efficiency and effective-ness of the drug development process.

Drug development is based on the requirement of theregulatory authorities that the sponsors of new drugs mustprovide “substantial evidence” of safety and effectiveness.The agencies generally require completion of three phasesof human trials to establish safety and efficacy of newdrugs, each one progressively expansive in scope, durationand cost [4]. The time and cost needed to develop a newcompound have increased considerably. DiMasi et al. [5]calculated that the average cost of bringing a drug to themarket increased from US $ 318 million in 1991 to US $

Wolf S. Richter ())Schering AG,Global Medical Development Diagnostics,Müllerstraβe 178,13342 Berlin, Germanye-mail: wolf-stefan.richter@schering.deTel.: +49-30-46818569, Fax: +49-30-46898569

European Journal of Nuclear Medicine and Molecular Imaging Vol. 33, No. 13, July 2006

802 million in 2003 (inflation adjusted, including oppor-tunity cost of capital). The cost calculation comprises theexpenses for failures of drug candidates in the developmentprocess. The average probability that a drug candidate willsuccessfully pass clinical phase I is in the range of 75%; therespective values for phases II and III are 50% and 65% [6].In total (including further probabilities, e.g. for the regu-latory review), the cumulative probability that a leadingdrug candidate will successfully proceed from the precli-nical phase to approval is about 8% (i.e. for every 12–13compounds that were serious candidates in preclinicalresearch, only one drug will make it onto the market) [6].

The rising cost of drug development is imposing asignificant burden on the pharmaceutical industry. Theattraction of integrating biomarkers into drug developmentincludes the expectation that less promising projects maybe stopped earlier (especially before they enter into costlyclinical phase III) and that the total cost of drug develop-ment will be reduced.

Personalised medicine

The progress in biosciences has advanced disease under-standing. The current classification of disease (which,especially in oncology, is preferentially based on anatom-ical criteria) will likely be broadened to encompass funct-ional/molecular parameters such as the expression patternand activity of different target molecules (e.g. HER-2/neuexpression). The confirmation of target expression andaccessibility is a prerequisite for the efficient selection ofpotential responders to target-specific therapies (“persona-lised medicine”). For this individualised diagnostic assess-ment, in vitro biomarker tests and in vivo imaging withspecific PET probes hold great promise. Since in oncolo-gical disease different lesions may show a different expres-sion profile even in the same patient [7], in vivo imagingmethods like PET are likely to be of superior value for acorrect classification in the future.

The transition to a focus on individualised medicine willbe mirrored in the drug development process. The integra-tion of biomarker results (in vitro tests and in vivo imaging)into clinical development programmes bears the potentialfor improved selection of patients for clinical trials, therebyincreasing the number of responders to (target-)specifictreatments. This will probably change current paradigms ofdrug development and transform and expand phase II trials,while potentially shrinking, shortening and reducing thecosts of phase III studies [8].

Biomarkers in early and late clinical development

The “Biomarkers Definitions Working Group” has elabo-rated a conceptual framework and preferred definitions forbiomarkers and surrogate endpoints [9] (see Table 1).

By logical extension, “imaging biomarkers” are anyanatomical, physiological, biochemical or molecular pa-rameter detectable by one or more imaging methods used toestablish the presence and/or severity of disease [4].

Both biomarkers and surrogate endpoints serve drugdevelopment purposes and may be used where they arevalidated for a particular application at issue [10]. Bio-markers have been categorised into three distinct categories(target, mechanism, clinical) on the basis of their contribu-tion to the logic of a clinical plan: first to confirm hitting thetarget and then to test two concepts, namely, that hitting thistarget alters the pathophysiological mechanism and thataltering this mechanism affects the clinical status [11].

In contrast to biomarkers (which are of specific value inearly drug development), surrogate endpoints have specificrelevance for the drug approval process of regulatoryagencies, e.g. the US Food and Drug Administration(FDA). In its 1992 accelerated approval regulations for newdrugs and biologics, it is stated that the “FDA may grantmarketing approval for a new drug product on the basis ofadequate and well-controlled clinical trials establishing thatthe drug product has an effect on a surrogate endpoint that is

Table 1. Definition of biomarkers, surrogate endpoints and clinical endpoints according to [9]

Biological marker (biomarker) Surrogate endpoint Clinical endpoint

A characteristic that is objectivelymeasured and evaluated as an indicator ofnormal biological processes, pathogenicprocesses or pharmacological responses toa therapeutic intervention.

A biomarker that is intended to substitute fora clinical endpoint. A surrogate endpoint isexpected to predict clinical benefit (or harmor lack of benefit or harm) based onepidemiological, therapeutic,pathophysiological or other scientificevidence.

A characteristic or variable that reflects how apatient feels, functions or survives.

Biomarkers may be useful in the assessmentof efficacy as well as safety.

Surrogate endpoints are a subsetof biomarkers.

Clinical endpoints are distinct measurementsor analyses of disease characteristics observedin a study or a clinical trial that reflect theeffect of a therapeutic intervention. Clinicalendpoints are the most credible characteristicsused in the assessment of the benefits andrisks of a therapeutic intervention inrandomised clinical trials.

S7

European Journal of Nuclear Medicine and Molecular Imaging Vol. 33, No. 13, July 2006

reasonably likely, based on epidemiologic, therapeutic,pathophysiological or other evidence, to predict clinicalbenefit or on the basis of an effect on a clinical endpointother than survival or irreversible morbidity...”, (US Codeof Federal Regulations 21CFR § 314.510 and 21CFR§601.419). In these regulations (and confirmed in the FDAModernization Act 1997) the use of surrogate endpoints inclinical trials is explicitly incorporated as a substitute forclinical endpoints.

The major difference between a biomarker and a sur-rogate endpoint is the level of validation. For a biomarker tobe accepted as a surrogate endpoint, considerable evidenceis usually needed from large and controlled clinical trialsabout the relationship to the true clinical endpoint. In manycases, the efforts required for the validation of a surrogateendpoint are beyond the capabilities of individual academicinstitutions or single pharmaceutical companies. Surrogateendpoints which have been accepted by regulatory author-ities for drug approval purposes in the (accelerated)approval context include the RECIST criteria for tumourresponse assessment in oncological trials (surrogate for theclinical endpoint: survival), lowering of cholesterol levels(clinical endpoint: cardiovascular events), increase in CD4count (clinical endpoint: improved survival in AIDS) andthe number of cerebral lesions on MR imaging (clinicalendpoint: disease progression in multiple sclerosis). Theobjective assessment of tumour response using diagnosticimaging became an accepted measure of cancer chemo-therapy after response criteria had been established on thebasis of bidimensional [12] and (more recently) unidimen-sional [13] tumour measurements.

Especially in oncology, the use of surrogate endpoints isof major relevance. For 71 oncology drug applicationsduring the 13-year period between 1990 and 2002, Johnsonet al. [14] evaluated the endpoints used as the approval basisby the FDA. In this survey, endpoints other than survivalwere the approval basis for 68% (39 of 57) of oncology drugmarketing applications granted regular approval and for all14 applications granted accelerated approval. Among thesurrogate endpoints, the assessment of tumour response wasmost frequently employed. The FDA has not acceptedchanges in tumour markers alone as a basis for marketingapproval of oncology drugs.

With regard to the use of 18F-FDG PET as an imagingbiomarker, the database for tumour response assessment isnot yet regarded as sufficient for employment of thismethod as a surrogate endpoint for approval purposes.

Early phase: translational medicine

Translational medicine (translational research) is concernedwith moving basic discoveries from preclinical researchinto clinical evaluation. The concept of translational medi-cine is of particular relevance in the framework of per-sonalised medicine. Specific challenges arise for target-specific compounds as animal models are not fully capableof mirroring the complete range of human pathology and,therefore, have only limited predictive value for efficacy

and safety in patient trials (e.g. the human target is notexpressed in the animal).

For the purposes of drug development, imaging withPET [and single-photon emission computed tomography(SPECT)] bears significant potential in translational med-icine as it allows the same methodology to be employed inanimal experiments (animal PET/SPECT scanner) andhuman trials. Using PET, basic concepts validated inanimals can be transferred to humans, potentially servingas imaging biomarkers for efficacy and/or safety in laterphases of the clinical development program.

The appeal of PET is also intimately linked to thediversity of potential radiotracers. In principle, theadministered drug can be radiolabelled directly to studybiodistribution and pharmacokinetics, or a different tracermay be employed for visualisation of the target itself or ofeffects of treatment [15]. The potential value for drugdevelopment includes (a) the assessment of the relativetargeting efficacy of different agents in patients for selectionof a lead compound for further development, (b) theassessment of human biodistribution and pharmacokineticsin comparison to respective values obtained in the animal,(c) the generation of supporting information for doseadjustments, (d) the assessment of drug availability at thetarget. Rudin and Weissleder [15] have summarised theprerequisites for imaging methods in translational medi-cine: the techniques have to be quantitative, reproducible,specific, sensitive, applicable to clinical practice and safe.PET is perfectly suited to meet these requirements.

Even though the scene is set for PET and other imagingmodalities to serve as major tools in translational medicine,reality is lagging behind the concept. The transition of newchemical entities from the lab into clinical studies withhuman subjects has not kept pace with the exponentialincrease in scientific understanding, a fact which has beenreferred to as the “translational bottleneck” [16].

During recent years, regulatory authorities have under-taken significant efforts to facilitate translational medicineapproaches. In June 2004 the European Agency for theEvaluation of Medicinal Products (EMEA) issued a “Po-sition Paper on Non-Clinical Safety Studies to SupportClinical Trials With a Single Microdose” [17]. In this paperthe requirements for safety testing were defined for com-pounds which are used in the context of translationalmedicine (including also PET). The clinical trials coveredby this position paper are exploratory in nature (pre-phase I)and may be conducted with a single test substance or with anumber of closely related pharmaceutical candidates tochoose the preferred candidate or formulation for furtherdevelopment. In any case the total amount of test compound(s) administered should not exceed 100 μg (and 1/100th ofthe dose calculated to yield a pharmacological effect). Themajor innovation of this position paper is that it regards asscientifically justified certain deviations from existingsafety standards in order to support pre-phase I trials and,thereby, to lower the barriers faced by investigators whenseeking early confirmation in humans. It is expected thatthis position paper will significantly facilitate translational

S8

European Journal of Nuclear Medicine and Molecular Imaging Vol. 33, No. 13, July 2006

medicine approaches and support innovation in the field ofpersonalised medicine.

A comparable situation exists in the USA, where, inJanuary 2006, the Center for Drug Evaluation and Research(CDER) of the FDA published a Guidance for Industry,Investigators, and Reviewers, which is comparable to theEuropean Microdosing regulations [18]. The FDA is cur-rently also discussing the manufacturing requirements forcompounds used in the context of exploratory (translatio-nal) clinical studies with special emphasis on radiolabelledPET tracers [19].

Late(r) phase: critical path

In addition to facilitating translational medicine approaches,the FDA has increasingly focussed during the past 2 years onlater clinical development phases. It is the belief of the FDAthat the “translational research efforts will not yield thehoped-for results without an analogous focus on downstreamdevelopment concerns.... A third type of scientific research isurgently needed, one that is complementary to basic andtranslational research, but focuses on providing new toolsand concepts for the medical product development pro-cess....”This highly targeted and pragmatic research is called“critical path research” because it directly supports thecritical path for product development success [2].

The urgency and need for critical path research is illus-trated by the apparent disconnect between medical practiceand regulatory requirements for clinical studies. 18F-FDGPET may serve as an illustrative example. In clinical rou-tine, 18F-FDG PET is increasingly used to assess the re-sponse to treatment in tumour patients; however, 18F-FDGPETas an imaging biomarker is not sufficiently validated tobe accepted as a surrogate endpoint for pivotal clinicalstudies by regulatory agencies.With this disconnect, patientflowwill need to be different in clinical studies as comparedto clinical routine—potentially impacting on the speed ofpatient recruitment and bringing into question the relevanceof the trial results.

With its critical path initiative the FDA proposes public–private collaborative work in applying technologies such asgenomics, proteomics, bioinformatics systems and newimaging technologies to the science of medical productdevelopment. These technologies are supposed to providetools to detect safety problems early, identify patients likelyto respond to therapy and lead to new clinical endpoints.Imaging technologies are specifically mentioned, but it isacknowledged that their predictive value needs furtherstudy and evaluation.

The validation of (imaging) biomarkers as surrogateendpoints is challenging. Criteria for validation include[20]: (a) relevance (a theoretical basis for their use in thespecific situation), (b) sensitivity and specificity for treat-ment effects, (c) reliability (accuracy, precision, robustness,reproducibility), (d) practicality (non-invasiveness, or mod-est invasiveness) and (e) simplicity (to facilitate widespreadacceptance). The most desirable paradigm for evaluation/validation of biomarkers is provided by adequate and well-

controlled clinical studies. It is the prospect of FDA’scritical path initiative that further (imaging) biomarkers bevalidated in a joint effort by public institutions, academiaand industry to serve as surrogate endpoints in clinicaltrials.

Requirements for effective use of imagingbiomarkers

Biomarkers serve different purposes in drug development.Biomarkers employed in early clinical development aresupposed to enhance product knowledge to further eluci-date the drug’s potential in different patient groups anddisease states. The information obtained at this stage ismainly important for designing subsequent clinical trialsand for internal decision making, i.e. to identify promisingdrug candidates and to stop the not-so-promising projectsearly. The level of validation needed for biomarkers in thisstage is determined by the degree of risk the pharmaceuticalcompany wants to accept. In many cases the necessaryvalidation will be confined to an assessment of intra- andinterindividual variability of test results in different relevantpatient groups. This type of “limited validation” will usu-ally be affordable for individual academic groups or singlepharmaceutical companies.

Biomarkers in later clinical development have the poten-tial to serve as surrogate endpoints and support approval formarketing authorisation. Validation as surrogate endpointsrequires significant resources and will, in many cases, onlybe possible in a concerted effort as proposed in the FDA’scritical path initiative. Considerations regarding the selec-tion of the best-suited biomarker in the context of a giventrial include practical aspects like the broad availability ofthe test method and the simplicity of its application. Tech-nically or logistically demanding procedures are, therefore,less likely to serve as effective biomarkers in later clinicaldevelopment.

The value of biomarkers will be greatest for target-specific compounds which are effective in a subgroup ofpatients only. Intelligent use of biomarkers will help toidentify in an extended clinical phase II programme thosepatient subgroups most likely to respond (or those patientsubgroups most likely to exhibit safety problems) and,thereby, to streamline and shorten the expensive clinicalphase III. Eventually, these biomarkers may serve as gate-keepers in phase III and later on the market. Intended use ofsuch biomarkers on the market necessitates a validation (orqualification for use) of the biomarkers themselves incombination with the therapeutic under consideration(which will expand the clinical trial programme).

The FDA has started discussions about the implicationsof a diagnostic–therapeutic co-development for regulatorypurposes and marketing authorisation. A preliminary con-cept paper is available from the FDA website [21]. In thisconcept paper the focus is on in vitro diagnostic tools;however, the basic considerations presumably also apply toin vivo imaging biomarkers.

S9

European Journal of Nuclear Medicine and Molecular Imaging Vol. 33, No. 13, July 2006

The implications of this development for the pharma-ceutical industry are currently not fully understood. Thera-peutic companies will need to consider the development ofdiagnostic tools, and diagnostic companies will have tostart to consider linking their new developments to thera-peutic applications.

Even though a more targeted use of therapeutics in thosepatient subgroups with the highest probability for efficacyand the lowest probability for side-effects according to theconcept of “personalised medicine” is desirable, this willpose significant challenges for the pharmaceutical industryas the size of the targeted patient group (and therefore themarket) gets increasingly smaller for a given drug candidate.This development has the potential to make it increasinglydifficult for pharmaceutical companies to recuperate the in-vestments involved in drug development during the life-timeof the product. To overcome such potential hurdles it will benecessary to commence a discussion between industry, aca-demia and regulatory authorities. Regulatory agencies havebeen successful in stimulating drug development for rarediseases (“orphan drugs”) and special populations in the past.Similar regulations might be necessary to provide a viableenvironment in order to realise the prospects of personalisedmedicine.

References

1. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC,Baldwin J, et al. Initial sequencing and analysis of the humangenome. Nature 2001;409:860–921

2. FDA. Innovation or stagnation. Challenge and opportunity onthe Critical Path to new medical products. 2004. http://www.fda.gov/oc/initiatives/criticalpath/whitepaper.html

3. Charles River Associates. Innovation in the pharmaceutical sector.A study undertaken for the European Commission. 2004. http://europa.eu.int/comm/enterprise/pharmaceuticals/pharmacos/docs/doc2004/nov/eu_pharma_innovation_25-11-04.pdf

4. Henderson JA, Smith JJ. Realizing the potential for biomarkersin imaging: background and legal basis. Food Drug Law J2005;60:511–524

5. DiMasi JA, Hansen RW, Grabowski HG. The price ofinnovation: new estimates of drug development costs. J HealthEcon 2003;22:151–185

6. Berndt ER, Gottschalk AHB, Strobeck MW. Opportunities forimproving the drug development process: results from a surveyof industry and FDA. MIT-FDA-Industry White Paper. 2006.http://www.nber.org/books/innovation6/berndt-et-al6-3-05.pdf

7. Shah RB, Mehra R, Chinnaiyan AM, Shen R, Ghosh D, ZhouM, et al. Androgen-independent prostate cancer is a heteroge-neous group of diseases: lessons from a rapid autopsy program.Cancer Res 2004;64:9209–9216

8. Fitzgerald GA. Opinion: anticipating change in drug develop-ment: the emerging era of translational medicine and therapeu-tics. Nature Rev Drug Discov 2005;4:815–818

9. Biomarkers Definitions Working Group. Biomarkers andsurrogate endpoints: preferred definitions and conceptualframework. Clin Pharmacol Ther 2001;69:89–95

10. Smith JJ, Sorensen AG, Thrall JH. Biomarkers in imaging:realizing radiology’s future. Radiology 2003;227:633–638

11. Frank R, Hargreaves R. Clinical biomarkers in drug discoveryand development. Nature Rev Drug Discov 2003;2:566–580

12. Miller AB, Hoogstraten B, Staquet M, Winkler A. Reportingresults of cancer treatment. Cancer 1981;47:207–214

13. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS,Rubinstein L, et al. New guidelines to evaluate the response totreatment in solid tumors. European Organization for Researchand Treatment of Cancer, National Cancer Institute of theUnited States, National Cancer Institute of Canada. J NatCancer Inst 2000;92:205–216

14. Johnson JR, Williams G, Pazdur R. End points and UnitedStates Food and Drug Administration approval of oncologydrugs. J Clin Oncol 2003;21:1404–1411

15. Rudin M, Weissleder R. Molecular imaging in drug discoveryand development. Nature Rev Drug Discov 2003;2:123–131

16. Hyman SE, Fenton WS. Medicine. What are the right targetsfor psychopharmacology? Science 2003;299:350–351

17. Committee for medicinal products for human use (CHMP),EMEA. Position paper on non-clinical safety studies to supportclinical trials with a single microdose. 2004. http://www.emea.eu.int/pdfs/human/swp/259902en.pdf

18. CDER, FDA. Guidance for industry, investigators, andreviewers—exploratory IND Studies. 2006. http://www.fda.gov/cder/guidance/7086fnl.htm

19. CDER, CBER, FDA. Guidance for Industry; INDs—ap-proaches to complying with cGMP during phase 1 (draftguidance, distributed for comment purposes). 2006. http://www.fda.gov/cder/guidance/6164dft.htm

20. Lesko LJ, Atkinson AJ Jr. Use of biomarkers and surrogateendpoints in drug development and regulatory decision making:criteria, validation, strategies. Annu Rev Pharmacol Toxicol2001;41:347–366

21. Department of Health and Human Services (HHS) F - Food andDrug Administration. Drug-Diagnostic Co-Development Con-cept Paper; draft - not for implementation. 2005. http://www.fda.gov/cder/genomics/pharmacoconceptfn.pdf

S10

European Journal of Nuclear Medicine and Molecular Imaging Vol. 33, No. 13, July 2006