Slide 1

Michelle GittelmanRutgers Business School

New Frontiers in the Economics and Management of InnovationKiTeS- Knowledge, Internationalization and Technology StudiesBocconi University, MilanMarch 22-23, 2012

The revolution that never arrived: Clinical and genetic paradigms in bio-medical discovery and the R&D productivity paradox1A puzzleThe pharmaceutical industry is having trouble filling its pipeline with new drugs despite doing many things right

Consistent increase in R&D expendituresMuch more basic science and genetics in drug discovery Increased use of analytical informatics Deeper division of innovative labor and active markets for technology, fueled by entrepreneurial firms spun off from universities

A puzzleThe pharmaceutical industry is having trouble filling its pipeline with new drugs despite doing many things right

ent increase in R&D expendituresMuch more basic science and genetics in drug discovery Increased use of analytical informatics eper division of innovative labor anmarkets for technology, fueled by entrepreneurial firms spun off from universities

The industry is doomed if we dont change - Chairman of Eli Lilly, 2007

This paperDescribe research paradigms and historical shifts in the institutional landscape of bio-medical researchWe study biotech as a revolutionary paradigm. What did it replace?

Genomics (1980s-present) locus of discovery is the labPatient-Oriented Clinical Research (POR) (1940s-1970s) locus of discovery is the hospital4Genomics discovery paradigm: Bench to bedside

DNA mutation associated with pathologyTargets in cellsDesign a drug to bind to targetClinical discovery paradigm: Bedside to bench

Observe perturbations in humans (e.g. unexpected reaction to a drug)

Test treatment experimentally

Theorize disease mechanismsIn the clinical paradigm, discovery starts with the whole organism. Perturbation can be unexpected reaction to a drug, this is used to then theorize mechanisms and treatments. There may be no theory of a cause mechanistic rather than theory driven5Competing discovery paradigmExperiential vs. theory-driven search

Rooted in ancient debates between Plato and Aristotle on the role of pure reason and experiential learning in advancing knowledgeKey issue in current debates in medical policy on translational research and systems biologyA central question in organizational learning literature (Arora & Gambardella, 1994; Gavetti and Levinthal, 2000; Nelson, 2003, Fleming and Sorenson, 2004)This paper: framing the issueCompare two research paradigms as different epistemologies of discovery different beliefs about the best way to find new medicinesDescribe the institutional history of clinical research in biomedicine in the USAReview the secondary evidence on discovery outcomesSuggest that productivity is linked to search paradigms much more research needed here!Search logicsExperientialAbstractType of search/Search routines8Search logicsTheory-driven, predictive logicFundamental cause-effect understandingsOffline experimentation

Abstract analytical modelsExperientialAbstractType of search/Search routines9Search logicsFeedback-based, backward lookingKnowledge of how objects workOnline experimentation

Real-world objects in real-world contexts

Theory-driven, predictive logicFundamental cause-effect understandingsOffline experimentation

Abstract analytical modelsExperientialAbstractType of search/Search routines10Search logics when problems are complexA striking characteristic of fields where technological advance has been rapid is that they all seem to be closely connected to a powerful applied science or engineering discipline

Nelson, 2003ExperientialAbstractType of search/Search routines11Search logics when problems are complexA striking characteristic of fields where technological advance has been rapid is that they all seem to be closely connected to a powerful applied science or engineering discipline

Nelson, 2003When complexity is high, theory-driven, predictive search yields better discovery outcomes than experiential learning

Arora & Gambardella (1994), Fleming & Sorenson (2004), Gavetti and Levinthal (2000)ExperientialAbstractType of search/Search routines12Search logics: framing genomicsTrial-and-errorBlind searchRandom screening of compoundsGenomicsRational drug designExperientialAbstractType of search/Search routines131980s/1990s: genomics was presented as a silver bullet in drug discoveryIn 1990, Congress approved ~$3 billion funding of the Human Genome Project to sequence the entire human genome with the promise that the knowledge would translate to a wave of new rationally designed drugsGenomics firms were founded to turn genetic information into drugs (Human Genome Sciences, Celera, Millenium, Incyte)Scientific entrepreneurship by star scientists core to the model (Zucker and Darby, 1998). The model attracted billions in funding from private investors and Wall Street hoping to capitalize on the promise of rational drug discoveryIncyte is based in Palo Alto, Calif., deep in Silicon Valley, and it is no coincidence that the heart of its headquarters is a vast, glass-enclosed room full of powerful computers. ''At the end of the day, it's the information that matters,'' said Randy Scott, the president and chief scientific officer. ''We are all about the application of Moore's Law to biology,'' he said -- a reference to the observation that computer processing power doubles every 18 months. Applying that exponential growth to genomics should produce similar gains for drug discovery, Dr. Scott said.

Incyte is based in Palo Alto, Calif., deep in Silicon Valley, and it is no coincidence that the heart of its headquarters is a vast, glass-enclosed room full of powerful computers. ''At the end of the day, it's the information that matters,'' said Randy Scott, the president and chief scientific officer. ''We are all about the application of Moore's Law to biology,'' he said -- a reference to the observation that computer processing power doubles every 18 months. Applying that exponential growth to genomics should produce similar gains for drug discovery, Dr. Scott said.

Death is a series of preventable diseasesWilliam Haseltine, Founder, Human Genome Sciences Search logics: framing genomicsTrial-and-errorBlind searchRandom screening of compoundsGenomicsRational drug designExperientialAbstractType of search/Search routines17Search logics: framing genomicsPatient-oriented Clinical Research (POR)GenomicsRational drug designExperientialAbstractType of search/Search routines18Patient-oriented clinical researchResearch performed by a scientist and a human subject working together, both being warm and alive (Schechter, 1998)Rejects the idea of disease causality as a useful starting point for drug discoveryCausal understanding is not useful in finding treatments.A dominant paradigm in bio-medicine in post-War USA, spurred by the federalization of research (NIH)Different predictive logics in science[T]here remains a real problem about the relevance of many model systems, and the inability of many to understand that in biology, unlike physics, we dont have great general laws or large forces operating that allow us to work from the bottom up in terms of clinical prediction

Rees, Jonathan. 2002. Two Cultures? J Am Acad Dermatol, 46:313-6.

20[T]here remains a real problem about the relevance of manymodel systems, and the inability of many to understand that in biology, unlike physics, we dont have great general laws or large forces operating that allow us to work from the bottom up in terms of clinical prediction

Different predictive logics in scienceThe great physicist-turned biologist Leo Szilard said that once he changed fields (no pun intended) he couldnt enjoy a long bath as he could when he could dream abstract physics in the bath. As a biologist he was always having to get out to check on some annoying little fact. It is the problem of predicting across several levels of biologic explanation, and the absence of the all encompassing general laws in biology, that accounts for the fact that most clinically relevant discoveries come from the clinic rather than the laboratory and not, contrary to what many believe, vice versa.

Rees, Jonathan. 2002. Two Cultures?

21[T]here remains a real problem about the relevance of manymodel systems, and the inability of many to understand that in biology, unlike physics, we dont have great general laws or large forces operating that allow us to work from the bottom up in terms of clinical prediction

Bedside-to-bench discoveries in medicineThe link between cholesterol and heart disease, which culminated in the development of statins in the 1980s, originated in experiments conducted in 1913, when the Russian scientist Nikolai Anichkov unexpectedly observed that rabbits fed high-fat diets developed atherosclerosis.The treatment for pernicious anemia was discovered from the mechanistic insight that feeding patients liver cured them the underlying vitamin deficiency (b12), identified decades later, was one of many complex causes22Bedside-to-bench discoveries in medicineObservations of surgical patients receiving a new sedative resulted in the unexpected finding of marked decreases in hallucinations and delusions among psychotic patients. The discovery of an effective treatmentfor psychosis subsequently facilitated new theories of brain activity associated with schizophrenia. Fundamental discoveries for the treatment of sickle-cell anemia were triggered by the bedside observations of clinical researchers, who noticed that some populations (infants and certain ethnic groups) showed irregularities in disease rates. Later discoveries of the underlying genetic manifestations of the disease were motivated by models developed through earlier clinical research. 23Early history: application of scientific principles to medicineIn the 19th century, medical education carried out in for-profit schools taught by practicing doctors with no scientific training1910: Two landmarksFlexner report - teaching-oriented medical schools, housed in universities, full-time university faculty. Medical education based on the European model.Rockefeller Institute Hospital founded to foster clinically-driven medical discoveryA blueprint for medical education emerged in the 20th centuryMedicine taught by practicing doctors with no research trainingFor-profit schoolsDebate over the modelApprenticeship in hospitals; for-profit private schools; research-oriented schools attached to universities.1910Flexner report recommended teaching-oriented medical schools, housed in universities, full-time university facultyRockefeller Institute Hospital founded to train scientifically-oriented doctors and give them a setting that combined bench and bedside to foster discovery By 1925, 80 medical schools were founded

train scientifically-oriented doctors and give them a setting that combined bench and bedside to foster discovery

24Organizing POR: Rockefeller Institute (1901) and Rockefeller Hospital (1910)First institution to combine laboratory and clinical work to find treatments for major infectious diseases of the day. Cosmopolitan, open culture, attracted top scientists from EuropeUnique scientific climate: Diverse specialization; transdisciplinary - no departmental divisions; minimal control by administrators

25Bedside to bench learning at RockefellerSimon Flexner, the first director of the Rockefeller Institute, conceived of the Rockefeller Hospital as a test site for the bright ideas generated in the Institutes laboratories. In fact, this has happened only rarely. During my 40 years at Rockefeller Hospital, I recall only one instance in which a laboratory observation by biochemists was turned into a testable hypothesis in patients. Indeed, the traffic of ideas often runs the other wayAhrens, Crisis in Clinical ResearchOrganizing POR: The NIH Clinical Center (1955) and GCRC networkModeled on the Rockefeller Institute 10x larger: 500 bedsVictory over science policy czar Vannevar Bush, who promoted government funding of basic research, not medical resaerchA model for a network of clinical sites in AMCs

27The bedside as the locus of discoveryScientifically, the most important asset of a POR facility is the golden opportunity it provides for medical investigators and their staffs to watch carefully and to think deeply about the medical challenges posed by their patients; this forces them to formulate new hypotheses and to devise new stratagems for attacking unsolved problems. There is time to ponder an unexpected event an unexplained turn in the course of the disease or a puzzling response to a medication and thus to obtain fresh insights into a disease or a manipulation under study. Ahrens, The Crisis in Clinical Research1970s-80s: Decline in clinical researchTwo (three) factors accounting for the decline in PORDeclining career opportunities for young PIsEmergence of genomics as a dominant discovery paradigmEroding institutional and financial support for PORChanges in healthcare delivery

Ahrens: Average hospital stays in GCRCs decreased from 16 days in the 1970s to 4-6 days in 199-0. Linking molecular and bedside research: designing a clinical research infrastructure. [Review] Robertson D. Tung CS. Journal of Molecular Medicine. 79(12):686-94, 2001 Dec.

In 1969 all studies were inpatient (outpatientstudies were not initially permitted on GCRCs) whereasby 1999 the balance had shifted fundamentally, with only18% of studies using the inpatient facility and 82%using the outpatient unit.The Vanderbilt GCRC has remained in the samephysical location throughout the period covered by thesedata, but Vanderbilt Medical School itself has grownenormously, and its faculty are accordingly more widelydistributed. This is reflected in the average distance ofinvestigators from the perimeter of the GCRC. In 1969the average investigators office was just 15 m from theunit, not much further away than a room in a contemporaryAmerican home. For many clinical scientists theGCRC was indeed their personal research laboratory,and it was located as conveniently to them as the typicalbench laboratory of a preclinical scientist. Thirty yearslater the expansion of Vanderbilts biomedical researchenterprise had entailed construction of several newbuildings around the original hospital building housingthe GCRC. Now the average distance to the GCRC hasincreased to 80 m, and for many users now the equivalentof one or two city blocks away. It is noteworthy thatthere is no possible relocation of the GCRC that wouldsubstantially reduce this average distance. This distancemay not be large by the standards of ordinary clinicalpractice, but it is an increase which constitutes a qualitativechange in the nature of the interaction of the investigatorwith the GCRC. Clearly there is no going back tothe days of the small academic medical centers, but suchfigures powerfully argue that the GCRC must becomegenuinely multicentric in the large and complex environmentof contemporary academic medical centers. Proximityto the primary work-site of investigators must receiveincreased attention in the future.

31Decline of POR: Career pressures32

The Promise of Biotechnology, ca. 1980Prior to the biotechnology revolution, molecular biology had no direct clinical applications1973: Recombinant DNA the isolation and transfer of DNA in normal genes allowed for a direct link between molecular biology and clinical applicationsShift in public opinion towards acceptance of the safety, ethics and promise of genetic researchRadically different institutional paradigmUniversity labs, not AMCsPatenting of genetic materials with unknown utilityStart-ups to commercialize innovations directly to the market

The Promise, ca. 2012Big PharmaBiotechGenomics and basic science in medicineNow acknowledged that genomics has been a bonanza for science but not for medicine. Few new drugs have emerged from the paradigm.Recent study at Brigham: 101 genetic markers that have been statistically linked to heart disease were shown to have no value in forecasting disease among 19,000 subjects followed for 12 years; a more valid predictor was the old-fashioned method of a family history Basic science and firm-level innovationGenomics as a business model has failed major firms sell diagnostics tests and home kits$100+ billion invested in the biotech industry never made money Growing empirical evidence that star scientists have a negative impact on firm-level innovation: nonstars and scientists in applied fields have a positive impact on performance in biopharmaceuticals (Baba et al, 2009; Breschi and Catalini, 2010, Gittelman and Kogut, 2003, Rothermel and Hess, 2007, Toole and Czarnitzky, 2009, Zucker and Darby, 2001)

36Concluding remarksScience is not homogenous enormous variation in search logicsExperimentation vs. theory-driven logicsReal world vs. reductionist methodsWe need to pay more attention to field-specific differences in explaining science-technology linksExperiential learning important for complex problemsAlmost NO empirical research on the clinical paradigm in medical discovery a rich terrain for future researchThank you! 1. Identify and define medical needsResearch on disease mechanisms3. Identify and validate targets involved in disease processesSearch for lead compounds that interact with targetOptimize the properties of the lead compounds to generate drug moleculesDrug development and pre-clinical studies (in vitro and in vivo)Genomics as a rational approach to drug discoveryRational design of molecules is gradually replacing random, trial-and-error experiments. . . Growth of scientific understanding in molecular biology and genetic engineering has clarified important aspects of human metabolism and the chemical and biological action of drugs. . . By studying the structure of receptors, scientists can design (typically on computer) a theoretical compound that matches a given receptor site, and is expected to counter a certain pathology. Arora and Gambardella, The changing technology of technological change, Research Policy, 23 (1994)40