生產力弔詭【蘇筠】

Author: Millie Yun SU- Assistant Professor (Education) in Singapore Management University.  Research topics on innovation management, qualitative research, knowledge boundaries.

 “The revolution re-visited: Clinical and genetics research paradigms and the productivity paradox in drug discovery” (Michelle Gittelman 2016)

Summary: Breakthroughs in genetics and molecular biology since the 1970s and 1980’s have been framed to revolutionize drug discovery and development.  This framing is a powerful force in the adoption of biotechnology in firms, universities, and public policy sectors.  However, if technology has experienced exponential advancement, shouldn’t we have more drugs being discovered and developed?  More than 40 years have passed, yet the productivity of new drugs has remained low, even drastic decline since the 1980’s.  The productivity decline is problematic and even paradoxical.  Has the biotechnology revolution never arrived as it promised? Or is it that the revolution not a revolution in the first place?  Michelle Gittelman’s paper reviewed two drug discovery paradigms and presented compelling historical evidence to argue that the revolution was bound to fail because the limitation of predictive science to solve complex problems in natural phenomena.

Clinical research paradigm vs. genetic science paradigm

Two distinctive paradigms in medical research: clinical research and genetic science paradigms represent an epistemic divide in the beliefs whether disease causality serves as the starting point of drug discovery.  Clinical research paradigm, mostly adopted in the 1940’s to 1970’s, is an experiential, feedback search process using human subjects or study objects as they exist in the natural world.  For clinical researchers, searching for new drugs does not require an understanding of the disease or identifying the cause of the disease, which is rather secondary.  Instead, mechanistic understanding of patients and their symptoms guide therapeutic search.  Therefore, real data about a drug’s properties in patients were obtained very early in the process, and medicinal chemists would use the data to further synthesize new compounds.    

In contrast with clinical research paradigm, genetic science paradigm believes that fundamental causality of diseases is central.  Drug discovery logic is organized by the belief that “genes cause diseases,” and that genetic and molecular information serve as the starting point of drug discovery.  Geneticists and molecular biologists analyze genetic information to identify targets; technologies such as high-throughput screening, combinatory chemistry and crystallography are used to generate “hits” on the target, and then biochemists and pharmacologists would develop compounds to bind to the target.  Drug discovery became a linear process, with testing of the compound in animals and particularly in humans come relatively late in the process.  As a consequence, a lot of drug discovery effort and investment could be wasted if the compound fails later in clinical trials.  

The revolution bound to fail

Although many have believed that science should guide technological search, as Gittelman pointed out, this argument is problematic because problems like drug discovery, is fundamentally complex, and predictive search with science knowledge is unfruitful to technological discovery.  In her words, “when natural phenomena are highly complex and variable, there is a considerable distance between the power of predictive rules and the unpredictable outcomes that emerge in variable states of nature” (p. 1573).  When scientists abstract cells and biological entities from natural settings to generate theories, there is a big distance between the abstract knowledge and applying that abstract knowledge in the natural world.  Scientists who want to advance targets from cells to clinical development will need to “re-contextualize” the target in real-world setting, which involves a great deal of experimental and procedural complexity.  As such, insights generated from the complex phenomena are mostly independent from predictive theories, and “ [stripping] out contextualized variables might actually reduce opportunity for discovery” (p. 1573). 

This problem of enforcing predictive science to search drugs is becoming evident in clinical trial results.  Recent studies have shown that failures of clinical trial are now happening in the later stages, like Phase II and III.  One study found that in 1991, failure of clinical trial was due to adverse movement and absorption of the drug, which occurs in Phase I.  However, by 2000, failures of clinical trials have shifted to toxicology and efficacy, which happen in Phase II and III, because they are either toxic or ineffective in treating the disease.  The adoption of biotechnology in analyzing genetics and molecular information has not decreased technological uncertainty but just shifted it to later stages, where significant development costs have been sunk.  Put differently, scientists have been working on ineffective drugs and only found out later rather than earlier.                

Implication to the drug discovery space: Not only is the linear model non-linear, but it is rather backward!  

Gittelman’s argument is compelling, because the problem of bridging basic and clinical sciences is evident in empirical work.  Both clinical and basic scientists found that the “gap is too big”.  As a consequence, both sides find it strenuous to “bridge the gap.”  In my own work, one infectious disease professor said. “it takes a lot of courage to (cross the gap),” meaning to translate academic invention to marketable products, or making the career switch from academia to entrepreneurship. However, the institutions and policies are set up in the way that it is impossible to adopt the clinical research paradigm again.  Current debates on ethical issues and regulatory infrastructure would not agree to drug discovery at bedside, which was one of the main reasons that led to the downfall of clinical research paradigm. 

Nevertheless, the conceptualization behind Gittelman’s paper is profound, because it essentially challenges the fundamental beliefs behind the current biomedical research policies.  Not only is the linear model non-linear, but it is rather backward! The underlying message from Gittelman’s argument is that we can no longer think in terms of bridging basic and clinical or translating academic science to industry science.  Instead, we need to see complexity of biomedical research as it is. As drug discovery needs to happen in natural setting, medical advancement should not take place in basic science, but happen in a setting considering disease phenomenon as it is.   

Implication to biotech policies: Lost in translation!

Many countries, from the U.S. to Singapore and Taiwan, have bought into the belief that fundamental or genetic science drives medical advancement.  Starting from the U.S., the National Institute of Health has been motivated to institutionalize “translational” research since 2005.  Singapore has also made the mistake by pouring billions of dollars in basic science hoping to produce new therapeutics and drugs within 20 years.  Following the U.S. footstep, both Taiwan and Singapore governments have also recently shifted to invest in translational research.  Both Taiwan and Singapore have also vowed to vitalize biotech industry with the goal to expedite innovation productivity.    

This kind of thinking is dangerous.  According to Gittelman and many others, translational model is flawed for three reasons.  First, translational model still adheres to the linear model of bench-to-bedside, with genetic science driving discovery and contextualized and refined by clinical science at a later stage.  Going down the translational path will only lead to the demise of ineffective drugs.  Second, the U.S. policy has adopted the model of using biotech entrepreneurs to “fill in the gap” between upstream academic science and downstream industrial application.  Between 1980s and 1990s, thousands of biotech firms were founded to commercialize scientific breakthrough. These firms were founded by highly trained university scientists who broker between universities and pharmaceutical industries.  But how many them have actually succeed or been acquired by large pharmaceuticals?  If biotech firms that have received billions of dollars and failed to translate science from basic to clinical, what makes universities and research institutions to believe that they will succeed?  Third, translational model also emphasizes on multidisciplinary teams, integrating diverse skills and knowledge.  However, we all know that forming multidisciplinary teams and getting them to work together are merely combining diverse specializations, but rather integrating and transcending differences in epistemic cultures, science practices and norms. 

Moving toward organizing complexity 2.0

It is unlikely that we will find a magic bullet solution, but the future is not dire, as we have known more about managing complex innovation than we used to.  In our research (Su and Dougherty 2018), we looked at specific scenarios converge basic and applied scientists and found that both clinical and basic scientists do share a common motivation to improve human health.  They share a common motivation to understand how complex science works in human, even though each of them has their own approaches and trajectories.  With different priorities, both have an intrinsic motivation to solve disease problems.  Our research found that their practices and knowledge join together when they realize how their questions complement one another.  They are also motivated to contribute their expertise to contextualize targets in clinical settings and participate in team works to demonstrate safety and efficacy patterns of drug candidates.  Even though practices and knowledge trajectories of the two paradigms evolve separately, there are conjunctions where they overlap and inform each other.  Out finding is not a solution, but it is a first step to better understand how to organize the two paradigms of drug discovery.    

Implication to management scholarship

Lastly, Gittelman’s work also speaks to our work in management scholarship.  A problem in the management is that business practitioners do not read our work and that our theories do not “translate” to practice well.  We also continue to buy into the belief that theories should drive solutions to complex real-world problems and that novelty comes from theory-driven research.  Both publication and the promotion systems are also the engine behind this problem.  Top-tier publications want to see papers with research questions motivated by theories and with answers to either validate or challenge existing theories.  However, management and innovation problems are complex and situated in real world.  Theories derived from abstract settings are removed from reality and lost in translation from theory to practice.  For management research to be taken seriously by scholars and practitioners, it is imperative to situate research in the real world and theorize the underlying phenomenon to advance fundamental understanding.   

Gittelman, M. (2016). The revolution re-visited: Clinical and genetics research paradigms and the productivity paradox in drug discovery. Research Policy, 45(8), 1570-1585.

Su and Dougherty (2018).  Knowledge convergence between basic and applied scientists for drug discovery innovation: A sociomaterial perspective.  Working paper.