The Case for Standards in Life Science Research Seizing Opportunities at a Time of Critical Need Research conducted by: About GBSI The Global Biological Standards Institute (GBSI) advances life science standards to enhance global health. Through policy initiatives, thought leadership, and education, GBSI seeks to expand the development and adoption of standards in support of credible, reproducible, and translatable outcomes in life science research. The Washington, D.C.-based institute provides an independent forum for government, non-governmental organizations, industry, academia, and other stakeholders to ensure that biological standards development is focused on priority needs. Table of Contents Contents Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Current State of Standards in Life Science Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Tackling Irreproducibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 Negative Effects of Irreproducibility: The Case for Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20 Life Science Research Standards: Adding Value and Accelerating Progress . . . . . . . . . . . . . . .26 Moving Forward to Solve the Problem of Irreproducibility . . . . . . . . . . . . . . . . .33 Appendix I: Supporting Materials . . . . . . . . . . . . . . . . . . . . . . .34 Appendix II: List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . .41 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 1 Foreword Reproducibility is the Foundation of Life Science Research. Reproducibility self-corrects for random observations, unconscious experimental bias and selective reporting, and poorly designed studies. Yet far too often, the inability to reproduce experimental data in the life sciences has resulted in the invalidation of research breakthroughs, retraction of published papers, abrupt discontinuation of studies, and reduced trust in the research enterprise. More alarming— to the scientific community, research funders, patient advocates, and the public at large—is that valuable time and resources are wasted by irreproducibility, and opportunities to enhance global health are delayed or lost. While the causes of irreproducibility are complex, we know from many sources, including this Report, that many of the causes can be traced to the absence of a unifying standards framework. The concept of standards is not new, nor specific to the life sciences. Standards have in fact served as the foundation of progress throughout millennia. Many advancements that we take for granted every day, including language, light bulbs, bridges, Wi-Fi, and the internet, would not be possible without the development and adoption of global standards within their respective industries. But unlike other fields of endeavor, life science research has very few broadly-implemented standards. It’s Time to Act. The global community can no longer afford the economic and intellectual drain that is caused by irreproducible research. GBSI and several independent organizations, including prestigious journals and the U.S. National Institutes of Health, have recognized the urgent need and are stepping up to address the issue. These early efforts must now be augmented with a much larger collaborative and unifying effort that engages and mobilizes all stakeholders across the life sciences to advance the development and adoption of standards. Everyone is affected by irreproducibility, from bench scientists to patients. Everyone is a stakeholder in these efforts. Everyone should have a voice. Everyone includes you. Together, let’s solve this urgent challenge. For more information or to learn how to get involved in community-wide efforts to effect meaningful change, I invite you to visit GBSI at www.gbsi.org. Leonard P. Freedman, Ph.D. President Global Biological Standards Institute 2 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Executive Summary In recent years, tremendous advances in the life sciences have improved the lives of millions of people. Breakthroughs in genomics and other technologies presage an unprecedented era of understanding human biology and disease. Recently, however, a rising tide of questions about the quality and reproducibility of research findings have arisen in scientific journals and in the lay press. Because the quality of scientific research is critical to both the life science community and the general public, the Global Biological Standards Institute (GBSI)—a multidisciplinary, non-profit organization dedicated to the advancement of life science (also known as biological) standards—commissioned this independent third-party evaluation of the life science research landscape. The objectives were to: • Evaluate the current state of research quality; • Identify areas of concern; • Highlight how the development and widespread adoption of standards could be beneficial to producing credible, reproducible, and translatable outcomes in life science research; and • Illuminate the case for standards. This Report is based on the results of nearly 60 in-depth interviews with prominent life science research stakeholders, including academic research leaders, biopharmaceutical R&D executives, government agency personnel, editors of peer-reviewed journals, and leaders of professional societies. In addition, an extensive literature review and an analysis of data from multiple sources, including the National Institutes of Health, National Science Foundation, and PhRMA, were performed. The Life Science Research Landscape Is Changing Traditionally, the quality of life science research has been maintained through the peer review process in the short-term and “self-correction” of erroneous hypotheses and results through additional studies over the longer-term. Recent changes in the life science research environment are putting pressure on these traditional systems and decreasing their effectiveness. These changes include: • Increasing complexity associated with growth of high-throughput, specialized technologies such as next generation sequencing and the emerging “omics”; • Expanding volume of publications in a resource-strapped environment; and • Growing focus on translating research findings into clinically-actionable results. Irreproducibility Is a Pervasive, Systemic Problem The evolution of the life science research landscape has led stakeholders to pay greater attention to the quality of research. In particular, there is increased awareness of and concern about irreproducibility of research findings. Studies in both commercial and academic settings show that irreproducibility is ubiquitous across settings. Similar experiences were shared by the life science research stakeholders interviewed for this Report. All life science research stakeholders—including academic laboratories and institutions, industry and industry associations, investors, government agencies, charitable foundations, journals, professional societies, standards development organizations, and policy and patient advocacy organizations— THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 3 Executive Summary need to confront the challenges that result from irreproducible research. Irreproducibility has extensive negative and lasting effects, including: • Inefficient use of time and resources across the life science research space, with high profile instances costing millions of dollars; • Damage to reputations of laboratories, institutions, journals, and funders; • Impairment of industry-academic relationships; and • Overall less favorable public opinion of life science research. Irreproducibility is a pervasive, systemic problem that has multiple direct and systems-level causes. These causes range from individual laboratory decisions, such as lack of an appropriate control for a particular experiment; to variability in journal publication requirements; to lack of consistent research quality systems. Fundamentally, irreproducibility stems from undefined variance in reagents, practices, and assays between laboratories. Standards Are the Solution Irreproducibility can be traced to an absence of a unifying standards framework. Biological standards are reference materials or consensus documents generated by professional communities to reduce differences in practices. Hence, standards provide an effective and feasible solution to problems stemming from uncontrolled variation. Standards have been successfully used in countless other fields, including other laboratory-based disciplines, to successfully reduce variability and improve quality and outcomes. However, few broadly-implemented standards exist in life science research. Because irreproducibility is highly prevalent and its effects are profound and lasting, all stakeholder groups interviewed for this Report agree that there is a need for additional standards in life science research. Multiple standards can be envisioned; for example, standards to ensure quality can be developed and implemented for reagents, assays, laboratory practices, and data analysis and reporting. The development and implementation of standards, whether as adherence to written documents or use of reference materials, requires community consensus and alignment around both the necessity for standards and their content. Ultimately, expanded adoption of life science standards will require: • Educational initiatives to raise stakeholder awareness of the purpose and benefits of biological standards and understanding of the standards development process; • Opportunities and forums for stakeholders to identify areas in the life sciences where accelerated standards adoption could provide maximum benefit; • Engagement of stakeholders with standards development organizations or material reference providers in the development of specific standards; and • Development of effective policies and practices within the life science research community to ensure the proactive development and periodic updating of biological standards. By increasing reproducibility, standards can benefit all stakeholders, counteract quality challenges, and be a unifying driver of current and future life science research quality improvement efforts. Uniting diverse life science research stakeholders in a standardization effort will be challenging. Nonetheless, influential entities within the life sciences community can all join in a global initiative to exchange ideas, identify areas of greatest need, and advance the development and adoption of biological standards. 4 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Introduction “Standard” is an inflammatory word for many biological scientists, one that conjures up images of bureaucracy, regulation, and paragraphs filled with obscure references and alphanumeric designations. In reality, life science (also known as biological) standards are either highly characterized reagents or written documents that outline community consensus around certain practices.1 Although government agencies can adopt specific standards as regulations, adherence to a majority of standards is voluntary.2 Standards have extensively benefitted many fields, from the railroad industry to clinical diagnostics, and standards underpin almost all the capabilities of modern life. However, although some life science research standards, such as those dealing with animal welfare, are broadly used, overall the adoption of standards in life science research has been limited. Are more life science standards necessary? The project culminating in this Report was undertaken to answer this question. Scientific research has entered a new realm of complexity. Researchers are discovering targets more quickly, and the increasing understanding of pathway biology has necessitated multidisciplinary approaches to “omics.” The environment is also competitive, with a strong drive to make and report new discoveries and to do so in a cost-effective way. There is a paramount need for each breakthrough to be a quality output that is also timely and economical. Over the past several years, concerns about the quality of research have garnered the attention of multiple stakeholders in the scientific community.3–6 Specifically, widespread and pervasive irreproducibility of research findings has been documented. As a consequence, millions of dollars of research funds have been wasted, and researchers must confront a small but growing negative public perception of scientists and the scientific method. Periodically, the scientific community has voiced concerns about irreproducibility, and some stakeholders have started to take action. However, thus far there has been no coordinated global effort to ameliorate these problems. The scientific community feels the profound effects of research irreproducibility and is ready to find a unifying solution to this problem. Standards can be the solution. Standards are a vehicle for reducing irreproducibility by aligning the community around consensus-based methods. Life science research stakeholders have highlighted the need for research standards as a way to increase reproducibility of research findings. Better understanding of where, how, and why standards can help the life science community in the future is one of the first steps to meet this need. The purpose of the research conducted to generate this Report is to evaluate the current state of the quality of life science R&D methodologies, identify areas of concern, and highlight why the development and widespread adoption of standards could be beneficial. The Report is a catalyst for action to inform and mobilize stakeholders, and to establish recommendations for enabling change. 1. ISO/IEC Guide 2:2004, definition 3.2. 2. American National Standards Institute (ANSI). [Internet] [cited 2013 Sep 17]. Available from http://www.ansi.org/about_ansi/faqs/faqs.aspx. 3. Ioannidis, J.P. PLOS Med. 2005; 2(8):e124. 4. Mobley, A. PLOS ONE. 2013; 8(5):e63221. 5. Westphal, S. “Unverified Science.” Boston Globe. 2013 Jul 8. 6. Jasny, B. et al. Science. 2011; 334(6060):1225. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 5 Methodology This summary Report, based on research commissioned by GBSI to an independent organization in 2013, reflects the opinions of multiple stakeholders and proposes broad-based solutions. The Report is based on an extensive research program comprised of broad secondary research and a detailed primary interview process. The secondary research program included a literature review and synthesis of over 100 peer-reviewed and other references from sources such as government agencies. Many of these are cited in the footnotes of this Report. Analysis of data from multiple sources—including the National Institutes of Health (NIH), National Science Foundation (NSF), and PhRMA—was performed. The primary interview component consisted of almost 60 in-depth, individual discussions with life science research stakeholders, including: • Academic research leaders; • Academic and industry researchers; Figure 1. Stakeholder Interviews N=57 • Biopharmaceutical company 60 research and development Experts in Standards (R&D) executives; Experts in Quality Management 50 • Editors of peer-reviewed Editors of Peer-Reviewed Journals professional societies; • Executives at charitable foundations that fund basic and translational research; Number Interviewed scientific and clinical journals; • Leaders of scientific Leaders of Scientific Professional Societies 40 Charitable Foundation Executives Academic & Industry Researchers 30 Government Agency Personnel Biopharmaceutical R&D Executives 20 • Government agency personnel; • Experts in quality management; 10 Academic Research Leaders and • Experts in standards. 0 Interviews were based on discussion guides developed for specific stakeholder groups. Interviewees had no knowledge Interviewed Stakeholders Note that in addition to stakeholders with exclusively journal editorial or professional society leadership responsibilities, multiple interviewees in the academic research leader group also serve in these capacities of the research sponsor during the interviews,7 and all interviewee comments remained anonymous. These candid discussions yielded insights that were analyzed both qualitatively and quantitatively. The findings of the secondary research program, analysis, and stakeholder feedback were assimilated into this summary Report. 7. After the conclusion of the interview, some interviewees chose to release their names to the interviewer and the sponsor without connection to any comments made during the interviews. In these cases, the sponsor’s name was revealed to the interviewee. 6 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Current State of Standards in Life Science Research Defining Standards Before a community can internalize issues surrounding standards in life science research, it needs to understand what “standards” are. Standards align people around a consensus-based method for performing important functions. Two commonly described categories of standards are materials standards and written consensus standards (Figure 2). Materials standards are highly characterized physical substances, such as chemical or biological reagents (known as reference materials or reference reagents), and are routinely used in several areas of life science research and healthcare. Written consensus standards are documents that outline community consensus around certain practices. Document (or “paper”) standards can cover a broad range of practices, such as designating names, defining common sizes, outlining processes or procedures, and establishing quality systems. Because irreproducibility is highly prevalent and its effects are profound and lasting, all stakeholder groups interviewed for this Report agree that there is a need for additional standards in life science research. Multiple standards can be envisioned; for example, standards to ensure quality can be developed and implemented for reagents, assays, laboratory practices, and data analysis and reporting. Figure 2. Definition and Types of Standards Standard: A document or biological material that serves as a guideline and is established through broad community consensus Materials Standards: Well-characterized, purified biological or chemical reference materials Reference Reagents Written Consensus Standards: Consensus documents describing optimal practices Specification Standards Analytic Standards • Used for assay validation and calibration or directly in research and development • State an agreed value for something and may specify materials standards used for calibration • Can specify assay methodology, cutoff values, and/or calibration reagents • Describe optimal procedures for performing a defined task • Example: Standard viral strains; size standards for mass spectrometry • Example: Meter is the length travelled by light during 1/299,792,458 of a second • Example: ANSI/ATCC ASN0001 standard provides a uniform methodology for assaying activities of anthrax toxins • Example: MIAME criteria for microarray data reporting; standards for ethical treatment of laboratory animals THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Procedural Standards Systems Standards • Establish systems to enable consistency of practices, including use of other standards • Example: ASQ or ISO 9000 quality management standards outline effective quality management systems 7 Current State of Standards in Life Science Research Although government agencies can adopt specific community standards as government regulations, thereby making them mandatory, adherence to standards is most frequently voluntary. Adherence to voluntary standards is commonly checked through accreditation or certification, a process during which an organization is inspected by an independent body to ensure that it is adhering to standards. Standards are broadly used in a variety of fields. Both materials and document standards have been extremely effective in improving reproducibility and quality of laboratory-based fields, such as blood banking and clinical laboratory diagnostics. The fact that standards have been a part of these evolving fields for many years has allowed for documentation of benefits that can be achieved through standards development and implementation over time.8 Current Standards in Life Science Research In life science research, however, standards are not as prolific. The most well-developed standards and accreditation system exist to protect laboratory animal welfare. In the United States, this system is centered on the Guide for the Care and Use of Laboratory Animals, a collection of standards for appropriate laboratory animal care, initially published in 1963 and revised multiple times thereafter. Adherence to the principles in the guide is required “Very few standards apply to biological research. It can be a significant problem, particularly for assays, because everyone is doing something different and it is difficult to compare results across studies.” for institutions that receive Public Health Service9 support (including grants from the NIH and CDC10), and is monitored internally by each institution through the self-regulating Institutional Animal Care and Use Committees.11 Academic Research Leader Materials standards are also used in some research communities, particularly in areas like vaccine and gene therapy development and microbiology. Some research communities have also developed standards that apply to specific fields, such as standards for characterizing and identifying human embryonic stem cells.12 But other than in such highly defined areas, few established and broadly-implemented standards exist in life science research. Is this a concern? The feedback from the scientific community interviewed for this Report suggests that it is. 8. For more examples of clinical standards, please see Appendix I. 9. Committee for the Update of the Guide for the Care and Use of Laboratory Animals. Guide for the Care and Use of Laboratory Animals, 8th Ed., NRC 2011. 10. Centers for Disease Control and Prevention (CDC). 11. Institutional Animal Care and Use Committee (IACUC). [Internet] [cited 2013 Aug 18]. Available from http://www.iacuc.org/aboutus.htm. 12. Loring, M.F. and Rao, M.S. Stem Cells. 2006; 24:145–150. 8 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility Changes in the Life Science Research Landscape Over the past 30 years, life science research advances have benefited the life and health of millions of people. For example, with treatment, HIV patients in the United States and Canada can expect to live an almost a normal life span of over 70 years.13 Routinely recommended genetic testing for diseases such as cystic fibrosis (CF) allows detection and estimation of disease risk prior to pregnancy, enabling couples to make informed decisions, and some CF-affected individuals to live until adulthood.14 These advances are the result of translating basic research findings into clinical treatments or laboratory tests. And importantly, over 50% of basic research is performed in the academic setting.15 The current practices used to maintain research quality are deeply entwined with the fundamental tenets of academic research culture and have successfully produced most biological and treatment advances that we enjoy today. However, changes in the life science landscape, including rising complexity, competition, economic challenges, and translational focus, increase the challenges associated with maintaining life science research quality through traditional systems. The general flow of the current academic life science research process is summarized in Figure 3. Figure 3. Academic Life Science Research Process Replication by other laboratories Peer review Generation of derivative results Pl review NIH ethics regulations Synthesis of Results Pl review Publication Manuscript Preparation Project Idea Grant review process Replication Obtain Funding Technology Transfer Pl review Data Analysis Journal requirements Journal requirements Statistician review Experimental Design Quality Checkpoints Institutional policies Broadly used Moderately used Rarely used Record Data Pl review Experimental Protocols Pl review Intralaboratory replication Pl review Perform Experiment Equipment Calibration Institutional review (animal experiments) Pl or lab manager review Reagent Validation Peer review Green circles indicate common steps in the life science research process. Adjacent color-coded text describes current/traditional quality checkpoints. 13. Hogg, R. Increases in Life Expectancy among Treated HIV-Positive Individuals in the United States and Canada, 2000–2007. IAS. 2013:TUPE 260. 14. American Congress of Obstetricians and Gynecologists (ACOG). Update on Carrier Screening for Cystic Fibrosis. 2011. 15. National Science Board. Science and Engineering Indicators 2012. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 9 Tackling Irreproducibility Currently, the maintenance of life science research quality relies on a set of key checkpoints (Figure 4). Although the process is iterative, there are currently few standards surrounding quality management in life science research. Figure 4. Common Academic Life Science Research Process Checkpoints Quality Checkpoint Description Research Findings • Grant proposals describe purpose of research, preliminary data, high level methodology for future experiments, and resources needed Grant Review Intralab PI Review • Proposals reviewed by a panel of scientific experts recruited by the funder using funder-defined criteria that may incorporate adherence to guidelines or standards • Informal one-on-one and group laboratory meetings with trainees and staff • Internal laboratory policies (generally informal) • Experts selected by journal determine if paper results are based on sound scientific practices and conclusions stem appropriately Peer from the data Review of Manuscript • Journal editors ensure paper adheres to journal policies that may incorporate guidelines or standards Colleague Replication • Published results can be replicated by other labs using methodology available in the publication • Only a fraction of all work is replicated • Conclusions of prior studies used to generate new hypotheses SelfCorrection • As new hypotheses are tested, the prior conclusions are either validated or refuted Scientific Knowledge Legend: High-Quality Work Low-Quality Work The funnel represents the overall research process, with circles representing high-quality (yellow) or low-quality (blue) work. As the work flows though the checkpoints described on the right, most of the low-quality research is filtered out. Lack of broadly-accepted standards results in extensive variability in the ways these quality checkpoints are “I guide my students but also try to give them freedom in experimental design. We meet informally every week to discuss the data; it’s an iterative process.” Academic Research Leader implemented. For example, academic research leaders and researchers report large differences between laboratories as to which steps of the research process are routinely monitored by the principal investigator (PI). Only a few steps, such as manuscript preparation, are carefully checked for quality. 10 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility Peer review of manuscripts prior to publication is a particularly powerful checkpoint that many stakeholders believe fundamentally protects research integrity. However, journals exhibit significant diversity in requirements for publication as well as in the criteria that different journals and reviewers “In my laboratory, we don’t let results go to publication use to evaluate work.16 Currently, few broadly implemented standards exist to help journals align without someone who was not involved in the around requirements for publication. Although some experiments replicating them. We have caught many journals may incorporate guidelines and standards errors this way over the years and experiments that were into publication decisions, many do not. just not robust enough to withstand scrutiny. However, Publications release and disseminate results into the I don’t see this being done in most laboratories.” broader scientific community where findings and Academic Research Leader interpretation face additional scrutiny. The feedback from the life science community suggests that researchers believe that when the journal peer review process fails and erroneous data or interpretations get published, highly important or controversial results will be replicated and either confirmed or disproved by other laboratories within a few years. If findings are reproducible, other “If one journal doesn’t accept a paper, there is always laboratories will use them as the basis for a different journal that you can send it to. Different additional research, resulting in further publications and the growth of the field. If journals look at different things. Some look at how findings are not reproducible, the situation is well the work was done, others want great work more variable. When the original hypothesis is that will also attract a lot of publicity and attention. of tremendous significance, the irreproducibility of results is usually published, although this Some will publish anything.” process can take several years. For results of less Academic Research Leader significance, the inability to replicate findings is frequently not discovered nor reported. In some cases, reporting guidelines for certain types of studies or laboratory procedures may be developed by a professional society, and adherence to guidelines may be required by its journal as a condition for publication. For example, Cytometry A and Cytometry B, the journals of the International “Discrepancies between laboratories can arise because Society for Advancement of Cytometry (ISAC), require adherence to the MiFlowCyt standard for minimum they may use different cut-offs for compensation, a data analysis technique frequently used in flow cytometry. Because publications rarely required reporting this level of detail, there was no way for the replicating laboratory to determine what values were used initially, resulting in irreproducibility. The MIFlowCyt standard prevents this.” Academic Research Leader and Journal Editor information required to report the experimental details of flow cytometry experiments.17 However, this type of standard is rare in life science research. The current research landscape is shaped by multiple changes, including novel technologies, increasing specialization, and imperatives to make a faster impact on patient outcomes. These changes are exerting pressure on traditional paradigms for maintaining research quality (Figure 5). 16. Bohannon, J. Science. 2013; 342(6154):60–65. 17. Cytometry Author Guidelines. [Internet] [cited 2013 Aug 18]. Available from http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1552-4930/homepage/ForAuthors.html. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 11 Tackling Irreproducibility Figure 5. Pressures on Traditional Academic Life Science Research Process Checkpoints Traditional Life Science Research Quality Check Points Research Findings Increasing Complexity and Collaboration Grant Review Growing Competition Intralab PI Review Peer Review of Manuscript Colleague Replication Economic Challenges SelfCorrection Greater Translational Focus Scientific Knowledge Legend: High-Quality Work Low-Quality Work The funnel represents the overall research process, with circles representing high-quality (yellow) or low-quality (blue) work. As the work flows though the checkpoints described on the right, most of the low-quality research is filtered out. As pressures represented by the blue arrows are applied to the process, the filters become more permeable, allowing more low-quality work to pass through. Complex and high-throughput technologies and “It used to be that only flow cytometrists did flow cytometry. Now everyone wants to have a flow figure in their paper. The researchers rely on the expertise of our core facility to guide them through the process. They have to take it on faith that the core facility is doing things correctly.” Academic Research Leader disciplines, such as genomics, proteomics, cellomics, metabolomics, and others, generate increasingly large amounts of biological data.18 For example, the Human Protein Reference Database grew from 2,750 proteins in 200319 to over 15,000 today.20 The complexity and breadth of techniques leads to more specialization and a greater reliance on core facilities21 among researchers as the intricacies of each area present greater challenges and require more focused expertise. 18. Vucic, E. et al. Genome Res. 2012; 22(2):188–195. 19. Peri, S. et al. Genome Res. 2003; 13(10):2363–2371. 20. Human Proteinpedia. [Internet] [cited 2013 Jul 26]. Available from www.humanproteinpedia.org. 21. Core facilities are centralized, usually institutional, laboratories dedicated to one or more complex techniques such as flow cytometry or microarrays, have the equipment necessary to perform these techniques, and employ experts in relevant methods and technologies. 12 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility Academic research leaders note that they are relying more on collaboration with Figure 6. Academic Research Funding and Output experts in complex technologies as a 110 towards team science is encouraged by 100 academic biomedical research in the United States.22,23,24 For example, the 2010 NIH Collaboration and Field Science: A Field Guide notes that “Innovations and advances that were not possible within one laboratory working in isolation are now emerging from collaborations and research teams that have harnessed techniques, approaches, and perspectives from multiple scientific disciplines and therapeutic areas.” At the same time, multimodal approaches and reliance on team science can undermine the individual investigator’s ability to 100% 90 80% 80 70 60% 60 50 40% 40 % NIH Grant Applications the NIH, which funds the majority of Number of PubMed Citations (10,000s) NIH Funding (USD B) routine part of their work. The trend 30 20% 20 10 0% 0 2005 2006 2007 2008 PubMed Citations (10000s) control the research process and make 2009 2010 2011 2012 NIH Research Grant Funding NIH Successful Grant Applications (%) peer review more challenging. With use of novel, complex technologies to address Sequester-driven and other budget cuts in 2013 will cause the NIH to lose ~$1.7B and the NSF to decrease new research grants by 1,000. a broader scope of biological and clinical problems through multimodal and team Data from PubMed.gov and NIH Research Portfolio Online Reporting Tool (RePORT). approaches, these scientists may not have expertise in all the methods being used to generate data for a particular project or paper. Detailed knowledge of experimental modalities and common pitfalls would allow the PI to maintain “I have several postdocs in my lab that trained in different fields than my expertise. They perform experiments that merge our areas together and help us work at the intersection. I can try to help them if the experiments are not going well, but ultimately I just don’t have the same level of understanding in these areas as I do in my field.” Academic Research Leader research quality by rapidly identifying trainee errors in experimental design or interpretation. However, this checkpoint is weakened when PIs must evaluate methods and technologies with which they are less familiar, including interpretation of data obtained from collaborators and core facilities or even routine use of kits. Similarly, journal reviewers selected for their expertise with a particular aspect of a paper might not be familiar with the intricacies of all the technologies used in the research, making it more challenging for reviewers to detect errors and inconsistencies. 22. NSF. Higher Education Research and Development: Fiscal Year 2011 Detailed Statistical Tables. NSF 13-325. 2013 Jul. 23. National Institutes of Health 2012 budget. [Internet] [cited 2013 Jul 26]. Available from http://www.nih.gov/about/budget.htm. 24. Federation of American Societies for Experimental Biology (FASEB). Federal Funding for Biomedical and Related Life Sciences Research, Fiscal Year 2014. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 13 Tackling Irreproducibility The academic research environment has become even more competitive. The percentage of successful NIH grant applications is declining (Figure 6) as NIH funding has flattened.25 That figure is likely to decrease further in response to 2013 budget cuts of approximately $1.6 billion.26 The NSF also plans to decrease new research grants by 1,000.27 Meanwhile, academic research output (e.g., the number of PubMed citations), is steadily increasing. Academic research leaders and researchers report “We like to approach the problem through a variety of methods and test the concept in multiple systems, but in the current environment this is becoming more difficult. I would like to wait and try the pressure from their institutions and funding agencies to produce more publications. These scientists feel that productivity expectations for advancing professionally and obtaining grants are increasing, particularly because professional advancement decisions frequently take concept in cells, several different animal models, and the quantity of published work into consideration. human materials and then publish a thorough paper The dynamic of flattened funding with increasing that really synthesizes the concept. Now, this would have to be three papers. If the third one doesn’t support the first, the first paper is still out there.” Academic Research Leader output leads to fewer available resources for each published paper. Research leaders believe that this resource scarcity leads to the publication of shorter, simpler papers that can be published more rapidly but may not thoroughly explore all aspects of a particular scientific problem, leaving material for subsequent publications. Additionally, the pressure to produce work faster might lead to less thorough experimental design than the PI may otherwise want. These factors can make PI control of research quality less effective. Life science research is also becoming more translational in nature. Although basic science “Ten years ago, my laboratory performed mainly research has the goal of advancing knowledge basic research. Now, we don’t have any projects about biologic processes, translational research without at least some translational focus.” focuses on producing knowledge and products— Academic Research Leader such as novel drugs or laboratory tests—that can lead to better diagnosis, treatment, prevention, and cures of human diseases. A growing emphasis on producing clinical results can shorten the time between basic research discoveries and use of the discoveries in patients during clinical trials. The increase in importance of translatability is driven in part by NIH funding priorities28 as well as pressures from institutions, disease foundations, and industry. These pressures are strongly felt by stakeholders. For example, research leaders report that many PIs are incorporating more translatable projects into their laboratories in response to the current NIH funding environment, leading some laboratories with primarily basic research experience to engage in translational research. Increasing translational focus is also manifested through a growing number of academic-industry collaborations,29 such as the California Institute for Biomedical Research (Calibr),30 Johnson & Johnson’s Innovation Centers,31 and Novartis and University of Pennsylvania Center for Advanced Cellular Therapy.32 Several academic organizations have also established institutes focused on drug discovery such as the 25. 26. 27. 28. 29. 30. 31. 32. 14 NIH Research Portfolio Online Reporting Tools (RePORT). [Cited 2013 Jul 8]. NIH. Fact Sheet: Impact of Sequestration on the National Institutes of Health. 2013 Jun 3. NSF. Notice No. 133. 2013 Feb 27. Waldman, M. Nature. 2010; 468:877. Schachter, B. Nat Biotechnol. 2012; 30(10):944–952. Timmerman, L. “Merck Returns to SD, Pouring $90M into New Schultz-Led Institute.” 2012 Mar 15. Johnson and Johnson. “Johnson & Johnson Announces Plans to Establish Innovation Centers.” 2012 Sep 18. Novartis, “Novartis and University of Pennsylvania Form Broad-Based R&D Alliance to Advance Novel T-Cell Immunotherapies to Treat Cancer.” 2012 6 Aug. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility Vanderbilt Center for Neuroscience Drug Discovery. These institutes combine academic expertise with structured industrial practices pertaining to processes and reagents to improve translatability and accelerate drug development. The growing translational focus of research makes it more challenging for sufficient time to pass for “self-correction” to take place before research is adopted by industry or used as a basis for clinical trials or off-label drug use. The shortening of the time available for self-correction as a result of institutional and patient pressures is highlighted by a recent controversy surrounding the reproducibility of results obtained when treating Alzheimer’s disease mouse models with a Food and Drug Administration (FDA)approved skin lymphoma drug Targretin® (bexarotene).33 Initial results released in Science in March of 201234 indicated that the drug might have promise in treating Alzheimer’s disease because it increases clearance of soluble AB amyloid, decreases the area of AB amyloid plaques, and reverses cognitive deficits in mouse models. A New England Journal of Medicine editorial35 at the time stated “A single report of this kind of preliminary evidence will require confirmation before Alzheimer’s disease investigators even consider launching clinical trials in humans,” and cautioned against off-label36 use of the drug in Alzheimer’s disease patients. Over the next year, multiple laboratories attempted to replicate this high-profile research.37-40 These studies, released in Science in May 2013, had more equivocal results. The data supported some but not all of the findings of the original work,41 dampening the enthusiasm many researchers felt about the drug’s promise. However, during the year between the publication of the original paper and subsequent publications, substantial activity to enable drug use in patients took place. A phase II clinical trial to study the use of bexarotene in Alzheimer’s patients was initiated in January of 2013.42 In April of 2013, Case Western Reserve University, where the original research was performed, granted an exclusive license for the treatment strategy to a spinoff company, ReXceptor. The company was co-founded by two of the original study authors and secured $1.4 million in funding from multiple sources, including Alzheimer's Drug Discovery Foundation (ADDF) and BrightFocus Foundation, to support a phase Ib trial in healthy volunteers.43 Furthermore, patient and family pressure to prescribe the drug off-label was significant,44 resulting in at least some use of a drug that has side effects but no proven benefit.45 The Targretin example illustrates the growing pressures that make it hard for scientific research to self-correct. Although replication of key data and discussions about interpretation are a routine part of the scientific process and enable self-correction of findings to take place, the time to allow this process to proceed before a drug is administered to patients has shortened. The example also demonstrates that the changes in the life science research environment are beginning to influence ways that research quality has been traditionally maintained. As stakeholders are affected by these trends, they are beginning to focus attention on quality indicators, such as the reproducibility of research findings. 33. Previously owned by Eisai, currently owned by Valeant Pharmaceuticals. Valeant Pharmaceuticals. “Valeant Pharmaceuticals Acquires U.S. Rights to Targretin® from Eisai Inc.” 2013 Feb 21. 34. Cramer, P. et al. Science. 2012; 335(6075):1503–1506. 35. Lowenthal, J. et al. N Engl J Med. 2012; 367(6):488–490. 36. Off-label use is prescription of a drug that is FDA-approved for one indication (e.g., skin lymphoma) to patients with another disease (e.g., Alzheimer’s). This practice is legal in the United States and frequently used in disease areas where few explicitly approved treatment options exist. 37. Fitz, N. et al. Science. 2013; 340(6135):924-c. 38. Price, A. et al. Science. 2013; 340(6135):924-d. 39. Tesseur, I. et al. Science. 2013; 340(6135):924-e. 40. Veeraraghavalu, K. et al. Science. 2013; 340(6135):924-f. 41. Shen, H. “Studies Cast Doubt on Cancer Drug as Alzheimer’s Treatment.” Nature News. 2013 May 23. 42. Bexarotene Amyloid Treatment for Alzheimer’s Disease (BEAT-AD) [Internet]. ClinicalTrials.gov [cited 2013 Jul 31]. Available from http://www.clinicaltrials.gov/ct2/archive/NCT01782742/. 43. PRNNewswire. “Research Foundations Collaborate to Fund Phase 1 Study of Cancer Drug in Alzheimer's Disease Patients.” 2013 Apr 24. 44. Wang, S. “Alzheimer’s Families Clamor for Drug.” The Wall Street Journal. 2012 Feb 11. 45. Koebler, J. “Study Finds ‘Serious Problems’ with 2012 Alzheimer’s Breakthrough.” U.S. News and World Report. 2013 May 23. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 15 Tackling Irreproducibility Reproducibility of Research Findings Increasing pressures on traditional systems for maintaining research quality are leading to an “If there is one retraction or irreproducible paper, increased awareness of irreproducibility of research findings among stakeholders. Reproducibility and the field forgives you. If there are retractions or reputation are key criteria that stakeholders use to irreproducible papers all the time, people will not judge the quality of scientific work (Figure 7). trust the journal. It means there is something wrong Reproducibility can be demonstrated directly when with your review process.” results are replicated externally or expected because Journal Editor of the intrinsic qualities of a paper.46 Number of Times Listed as Top 3* Way to Assess Research Quality Figure 7. Stakeholder Assessment of Quality N=40 90 80 Other 70 Impact on Field 60 Reputation 50 40 Externally Reproducible 30 Intrinsic Aspects of Paper Good Reputation • Internal consistency • Pl • Institution • Journal • Consistency across experimental systems • Appropriate data analytic techniques 20 10 • Strong experimental design Intrinsic Aspects of Paper Externally Reproducible Impact on Field • Importance • Novelty • Non-selective presentation of results 0 Responses to Open-ended Question “How do you define life science quality?” Stakeholders commonly use these connected features to assess the quality of life science research. Strong experimental design includes validation of reagents; appropriate selection of controls and experimental systems; blinded experimentation; and robust number of replicates. Bar graph shows number of times interviewed stakeholders noted a feature of life science research as a top 3 way to assess quality (N=40). *Some stakeholders reported only one or two top ways to assess research quality. Reputation is also viewed as a prominent indicator of “When you work in the field, you know what labs are good and which ones have data that you need to take with a grain of salt because you will not be able to replicate it.” Academic Research Leader quality. This criterion can include the reputation of the PI or laboratory that has produced the work, the institution where the research was performed, or the journal in which the work was published. Over time, the reputations of laboratories, journals, and PIs are mainly derived from producing work that is reproducible and stands the test of self-correction. 46. Begley, C.G. Nature. 2013; 497(7450):433–434. 16 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility Extent of Irreproducibility Irreproducibility is a pervasive, systemic problem experienced by most researchers across life science settings. Discussion about the irreproducibility of many studies, particularly those in genetics, has been ongoing since the early 2000s. This issue was highlighted in a seminal 2005 essay by John Ioannidis, a professor at the Stanford School of Medicine, titled “Why Most Published Research “We’ve had to address issues with replicating published work many times. Sometimes it’s a technical issue, sometimes a reagent issue, sometimes it’s that the technique was not being used appropriately.” Findings Are False.”47 In that paper, Ioannidis performed mathematical modeling and suggested that “for most study designs and settings, it is more likely for a research claim to be false than true.” The article postulated unconscious bias as a key cause. Academic Research Leader Several years later, two highly publicized experimental studies showed a significant lack of reproducibility of academic research findings in the industrial setting. In a 2011 Nature Reviews Drug Discovery article,48 Prinz and colleagues from Bayer Healthcare reported that in 67 projects over 4 years, published data was reproducible only 21% to 32% of the time, depending on the definition of reproducibility (all data vs. main dataset) used. A subsequent report coauthored by Begley and Ellis49 reported a similar experience at Amgen. Out of 53 “landmark” papers that the hematology and oncology department attempted to replicate over 10 years, only 11% were reproducible. Industry stakeholders interviewed “Our lab is focused on translating basic research for this study also report extensive experiences discoveries and we are only able to replicate 1 in 5 with lack of reproducibility. concepts that have passed through our lab.” Irreproducibility is not limited to the industrial Government Research Leader setting. A recent MD Anderson survey showed 50 that over 65% of senior academic faculty has had the experience of being unable to reproduce a finding from a published paper. Most tried to contact the original authors, but over 60% of the time they received an indifferent or negative response or no response at all. Among the stakeholders interviewed for this work, experiences with irreproducibility were reported by most academic research leaders, government research leaders, industry leaders, and journal editors. Over 80% of interviewed academic research leaders have had some experience with irreproducibility, either when trying to reproduce the work of others or having their own work reproduced (Figure 7). “Because we carefully select the labs we work with, in our hands, irreproducibility is lower than the Begley [2012] study, maybe 30%, but this is still very significant.” Industry Research Executive Causes of Irreproducibility Irreproducibility is the result of differences in performance of a particular experiment between laboratories. These differences can occur when one laboratory performs the experiment in a more optimal way, or even when both laboratories perform 47. 48. 49. 50. Ioannidis, J.P. PLOS Med. 2005; 2(8):e124. Prinz, F. et al. Nat Rev Drug Discov. 2011; 10 (9):712. Begley, C.G. and Ellis, L.M. Nature. 2012; 483(7391):531–533. Mobley, A. et al. PLOS ONE. 2013; 8(5):e63221. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 17 Tackling Irreproducibility the experiment optimally but with inherent differences in methods or reagents. Multiple systemic causes contribute to variability of both practices to ensure quality and performance of the specific experiment in question, including absence of formal or consistent research quality systems, variable education of staff, and differences in journal review and reporting policies (Figure 8). Many of these causes of irreproducibility can ultimately be traced to an underlying lack of a standards framework. Figure 8. Causes of Irreproducibility Experimental Irreproducibility Academic Experience Resource Limitations Reagent Variability/Lack of Validation Experimental Bias Increasing Complexity Differences in Analytic Techniques Differences in Statistical Methods No Quality Systems Unintentional Selective Reporting Lack of Appropriate Controls Variable Education Lack of Equipment Calibration Variations in Interpretation Scientific Misconduct Variable Journal Policies Legend: ➡ Pervasive ➡ Moderate ➡ ➡ ➡ ➡ ➡ ➡ ➡ ➡ ➡ ➡ ➡ % Senior Faculty Incomplete Protocol Reporting 75% 50% 25% 0% GBSI Mobley No Irreproducibility Experience Experienced Irreproducibility Industry Experience 100% % Papers Examined Lack of Standards Pressure to Publish 100% 75% 50% 25% Rare 0% Prinz Begley Results Confirmed Results Not Confirmed Academic experience refers to personal experience of academic senior faculty with irreproducibility. Industry experience refers to percentage of papers studies reported being able to reproduce. Graphs generated based on responses from current study interviews (labeled GBSI, N=16) and Mobley PLOS ONE (2013) (N=148). In industry experience, papers were considered reproducible when literature data was in line with in-house data. Graphs generated based on Prinz et al. in Nature Reviews (2011) and Begley and Ellis in Nature (2012). Causes of irreproducibility defined and ranked based on current interviews, Loscalzo in Circulation (2012) and Begley in Nature (2013). 18 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Tackling Irreproducibility These systemic causes translate into a variety of immediate causes of irreproducibility, including the following: • Incomplete protocol reporting; for example, when authors leave out certain details that they do not believe to be crucial to the performance of an experiment or that were never recorded as a formal part of the laboratory’s procedure for performing the experiment. • Reagent variability, such as can occur when two laboratories use what they perceive to be the same reagent, but are in fact two biologically-distinct forms.51 For example, significant differences can exist between antibodies52 to the same antigen from different manufacturers or even from different lots produced by the same manufacturer. • Lack of reagent validation, such as using unauthenticated cell lines (see “Life Science Standards at Work: Cell Line Authentication” box, page 28). • Experimental bias,53 such as unconsciously and unintentionally measuring the size of a tumor in an animal from the group an investigator knows to be treated as smaller than in the untreated group. Unblinded experiments where an investigator knows which cell culture dish or group of animals has received treatment are most prone to this type of bias.54 • Differences in analytic techniques; for example, using different assays (tests), imaging, or measurement techniques to evaluate experimental results. • Differences in statistical methods, including use of different mathematical approaches to analyze data or use of statistical approaches that might not be optimal for the particular data type. • Unintentional selective reporting, such as assuming that the experiment worked when results are positive but did not work when results are negative. • Lack of appropriate experimental controls, such as excluding a true negative control to understand the specificity of an antibody or a primer for detecting the antigen or sequence of interest. • Lack of equipment calibration leading to differences in measurements between laboratories. • Variations in interpretation, such as interpretation of a very weak band on a gel as positive by one laboratory and negative by another. • Scientific misconduct. Scientific misconduct has recently received attention in the scientific literature,55 the National Academy of Sciences,56 the 2013 International Congress on Peer Review and Biomedical Publication (Peer Review Congress),57 as well as in the popular press.58,59 Particular attention has been brought to the increasing number of peer-reviewed paper retractions and the fact that the majority of retractions result from scientific misconduct.60 Although there is limited overlap, irreproducibility is not equivalent to paper withdrawals or scientific misconduct. Only 0.01% of papers are retracted, whereas irreproducibility is a pervasive, complex issue that is far more prevalent than paper withdrawals and associated misconduct. Furthermore, most irreproducible papers are never withdrawn. Although scientific misconduct is certainly an issue of grave concern, most interviewed stakeholders view it as an uncommon cause of irreproducibility in the life sciences. 51. Vasilevsky, N. et al. PeerJ. 2013; 1:e148. 52. Antibodies are proteins produced by immune system cells that have a specific affinity to parts of other biologic molecules (antigens). Antibodies can be produced commercially, labeled with dyes, and used to detect specific biologic molecules (e.g., a particular protein) in scientific experiments. 53. Loscalzo, J. Circulation. 2012; 125(10):1211–1214. 54. Vesterinen, H.M. et al. Mult Scler. 2010; 16(9):1044–1055. 55. Sliwa, J. “Has Modern Science Become Dysfunctional.” American Society of Microbiology, 2012 Mar 27. 56. Fang, F.C. Proc Natl Acad Sci USA. 2012; 109(42):17028–17033. 57. Peer Review Congress. [Internet] [cited 2013 Sep 15]. Available from http://www.peerreviewcongress.org/2013/Plenary-Session-Abstracts-9-8.pdf. 58. Zimmer, C. “Misconduct Widespread in Retracted Science Papers, Study Finds.” New York Times. 2012 Oct 1. 59. Johnson, C. “Harvard Investigation of Stem Cell Scientific Misconduct Provides Insight into Secretive Process.” Boston Globe. 2013 Apr 8. 60. Steen, R. J Med Ethics. 2011; 37:249–253. 61. Corbyn, Z. “Misconduct Is the Main Cause of Life-Sciences Retractions.” Nature News. 2012 Oct 1. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 19 Negative Effects of Irreproducibility: The Case for Standards Irreproducibility Reflects the Need for Life Science Standards In its essence, irreproducibility is the result of frequently unrecognized differences in the performance of a particular experiment between laboratories or researchers. Solutions to this pervasive problem can focus on two areas: (1) increasing recognition of intentional differences between experiments conducted by different laboratories (e.g., better data reporting practices), and (2) reducing unintentional differences between laboratories. Standards can effectively reduce differences in practices by aligning the community around consensus-based methods, and are therefore an effective solution for irreproducibility. Multiple laboratory-based fields, such as clinical diagnostics and blood banking, have been able to successfully improve reproducibility and quality through standards. For example, standardization of HbA1c testing in diabetes significantly decreased variability of results between laboratories62 and the AABB (previously known as the American Association of Blood Banks) standard for bacterial testing of platelets resulted in a 70% decrease in transfusion-related sepsis63 (please see Appendix I for further details of both examples). Because of the significant consequences of irreproducibility, all stakeholders report the need for more life science standards (Figure 9). Figure 9. Need for Life Science Standards Need for Standards to: Improve Consistency of Reagents, Substrates, and Analytic Methods Improve Consistency of Daily Laboratory Processes Improve Experimental Design and Analysis Improve Sharing of Data Stakeholder Group Academic Researchers Industry Government Agencies Non-Profit Funders Journals and Professional Societies Examples of Possible Standards Antibody specificity, endotoxin positivity, methods for analyte measurement Legend: Laboratory data recording, maintenance of SOPs, quality systems Low Unmet Need Better procedure Minimum replicates, reporting, compatibility data analytic of dataset formats techniques, quality and labeling systems Some Unmet Need High Unmet Need Rankings are based on qualitative interviewee responses (N=50). Orange circles indicate that the majority of interviewed stakeholders reported no or very low need for additional biological standards. Light green circles indicate that the majority of interviewed stakeholders felt there is some need for additional standards. Dark blue circles indicate that the majority of interviewed stakeholders reported a prominent need for additional life science standards. 62. Little, R. Clin Chem. 2011; 57(2):205–214. 63. AABB. “Public Conference – Secondary Bacterial Screening of Platelet Components, 7/17/12.” 20 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Negative Effects of Irreproducibility: The Case for Standards Effects of Irreproducibility Research practice variability and quality challenges signified by irreproducibility can lead to inefficient use of time and resources across the life science research space (Figure 10). Figure 10. Effects of Irreproducibility Irreproducibility Academic Labs & Institutions • Increased time and effort to generate results and publications • Reputation impact Industry & Investors • Target validation and development failures • Unpredictable outcomes • Suboptimal use of funds • Failed investment Funders • Suboptimal use of financial resources • Delays in research progress • Reputation impact • Worse public perception Journals & Professional Societies • Delays in research progress • Reputation impact • Worse public perception Inefficient Use of Resources and Time Impact on Reputation Less Favorable Public Opinion of Research Irreproducible Findings Waste Academic and Public Health Resources Recent controversy surrounding the role of a retrovirus called xenotropic murine leukaemia virusrelated virus (XMRV) in causing prostate cancer and chronic fatigue syndrome (CFS) is an example of how extensive resources may be needed to show that irreproducible results are invalid (see call-out box on next page). THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 21 Negative Effects of Irreproducibility: The Case for Standards Irreproducible “Discovery” of XMRV XMRV was initially identified in human prostate cancer samples in 2006.64 In 2009, another study showed that XMRV infection is associated with higher grade of prostate cancer.65 Within a few months, a widely publicized study published in Science demonstrated that XMRV could be identified in almost 70% of patients with chronic fatigue syndrome (CFS), but in less than 5% of healthy volunteers.66 Due to the presence of a virus that might be a cause of disease and potentially transmissible through blood transfusion, the public health community began to voice serious concern about the need to protect the blood supply from XMRV. The AABB and U.S. government (through a collaboration between the NIH, FDA, CDC, and several academic groups)67 established groups to evaluate the potential risk to the blood supply. Multiple blood banks around the world began to refuse blood donations from patients with CFS.68 Given the potential consequences to the blood supply, multiple attempts to replicate the findings were made in Europe and the United States. Although one study supported the findings in CFS patients,69 multiple other studies did not find the virus either in CFS patients, prostate cancer patients, or in healthy controls.70 In 2010 and 2011, contamination of samples and laboratory reagents (such as commercial RT-PCR71 reagents) with XMRV derived from experimental mice or mouse cell lines was proposed as the cause of the initial positive results. Many additional studies, including two large multi-center efforts72.73 and studies examining thousands of donor blood units performed between 2010 and 2012, did not identify XMRV in human samples. It is now believed that the initial research findings were an erroneous result of contamination that was not detected due to lack of appropriate controls.70,72 Over the course of almost four years, and through the collective efforts of multiple laboratories, government agencies, blood transfusion services, and organizations around the world, the initial errors were corrected. However, substantial costs were associated with these efforts. Although the total research expenditure of dozens of laboratories and public health efforts by blood transfusion services around the world have not been calculated, the cost of one of the multi-center studies alone was approximately $2.3 million,72 suggesting that the overall costs were very significant. The potential public health implications make the XMRV example particularly dramatic, but even when the consequences are less compelling than blood supply safety, significant resources can be invested in trying to act on or replicate work that is inherently irreproducible. Academic laboratories can expend significant time and funds attempting unsuccessfully to reproduce work that they want to use as the basis for future experiments. This expense creates challenges for the laboratories and affects government and non-profit research funders of the original or subsequent research. Moreover, limited financial resources may be wasted on work that cannot lead to further advances. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 22 Urisman, A. et al. PLOS Pathog. 2006; 2(3):e25. Schlaberg, R. et al. Proc Natl Acad Sci USA 2009; 106:16351–16356. Lombardi, V. et al. Science. 2009; 326:585–589. Simmons, G. et al. Transfusion. 2011; 51(3):643–653. Kakisi, O.K. et al. Transfusion Med. 2013; 23(3):142–151. Lo, S.C. et al. Proc Natl Acad Sci USA. 2010; 107(36):15874–15879. Groom, H.C. and Bishop, K.N. J Gen Virol. 2012; 93(Pt 5):915–924. Reverse transcription polymerase chain reaction. Simmons, G. et al. Science. 2011; 334:814–817. Alter, H.J. et al. Mol Biol. 2012; 3:266–312. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Negative Effects of Irreproducibility: The Case for Standards A researcher recalls that as a graduate student he read a paper in a prestigious journal showing that a particular mutation could lead to protein interactions that cause carcinogenesis in several cell lines. The researcher generated several exciting hypotheses based on this Irreproducibility Complicates Academic-Industrial Relationships The inability to reproduce results also presents a significant challenge for industry. Biopharmaceutical companies might waste resources in attempts to optimize published academic procedures that are not work. However, his principal investigator insisted that replicable and therefore not translatable, as demonstrated because the findings in the paper were novel, the in the Amgen and Bayer experiences described above. researcher should replicate several of the key experiments before proceeding. The researcher Irreproducibility contributes to the unpredictability of development outcomes and failures in target validation and development. While most companies attempted to replicate the experiments for several months. After encountering multiple failures, he contacted the authors of the original paper. The authors were not happy about his inquiries, but agreed to help him. For several more months the researcher attempted to replicate the experiments, finally traveling to the original lab. When he was there, the authors could not replicate their published experiments at their own are highly protective of information regarding costs associated with these failures, one example is provided by Bruce Booth in his blog, LifeSciVC.com: “The company spent $5 million or so trying to validate a platform that didn’t exist. When they tried to directly repeat the academic founder’s data, it never worked.… Sadly this ‘failure to repeat’ happens more often than we’d like to believe. It has happened to us at Atlas several times in the past decade.”74 laboratory. The precise cause of irreproducibility was never pinpointed, but the researcher believes it may “Academic collaboration is the main way to reach key have been due to over-passaging of the cell lines used opinion leaders, particularly in areas where we don’t have by the original laboratory. Investing time in trying to internal expertise. Their experience and support is replicate these results delayed the researcher’s essential to our drug development efforts.” graduation by six months and slowed down his lab’s Biopharma R&D Executive progress on several other projects. He states: “It would have been wiser to quit earlier. But when the results are profound, you want them to be true and you want to make the next leap. Once you start trying to replicate something, it is difficult to say ‘This is just not valid. I will give up.’ You keep thinking that maybe it is just something you are doing wrong.” Academic Researcher Irreproducibility can significantly complicate academic-industry relationships at a time when both sides are finding these collaborations increasingly desirable.29 Government funding pressures are increasing the importance and desirability of industry collaborations for both institutions and individual laboratories. Academic research leaders report that in an environment of government funding challenges, industry collaboration can be a highly valuable source of funds and add stability to an otherwise volatile “We spent several months trying to reproduce findings in an animal model from another lab but couldn’t. That time and money was just lost.” Academic Researcher funding environment. Industry is also steadily seeking academic collaborations. Whereas early research and development investment in many companies has flattened, the number of commercial Investigational New Drug (IND) applications has declined (Figure 11), suggesting decreased R&D efficiency. 74. Booth, B. Academic Bias & Biotech Failures. LifeSciVC.com. 2011 Mar. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 23 Negative Effects of Irreproducibility: The Case for Standards Figure 11. Industry Early R&D Spend and Output To combat these trends, biopharmaceutical companies are increasingly focused $1.4 1,000 2005–2011 FL to I in industry on improving the translatability of $1.2 of otherwise unobtainable expertise and biological models for biopharma, resulting in a steady stream of collaborative activity. Collaboration with academic laboratories allows biopharmaceutical companies to access leaders in particular scientific disciplines and other experts that might not be available internally. Academic 800 $1.0 $0.8 600 $0.6 400 $0.4 $0.2 Commercial IND Applications opportunities. Academia provides a source Representative Discovery and Preclinical Research Spend (USD B) discoveries and pursuing commercializable 200 $0.0 laboratories can also provide valuable resources, such as cell and animal models 2005 approaches to quality can be challenging. Generally, while academic laboratories are 2007 2008 2009 2010 2011 Commercial IND Applications industry relationship is not without between the academic and industrial 2006 Representative Discovery and Preclinical Research Spend Although highly beneficial, the academicimpediments; mediating differences 0 $0.0 and active compounds. Representative R&D spend based on top four pharmaceutical (J&J, Pfizer, Novartis, Roche) and biotechnology (Novo Nordisk, Amgen, Gilead, Biogen) companies by 2012 revenue estimated based on company 10-K R&D spend and PhRMA annual member surveys, PhRMA Profile Biopharmaceutical Research Industry 2007–2013. Commercial IND application numbers based on Parexel 2012/2013. focused more on obtaining publishable results, industrial laboratories are more concerned with discovery that occurs Figure 12. Formal Quality Systems within the bounds of replicable processes 100% and can ultimately be transferred into the Have Formal Quality System clinical setting. Whereas maintaining quality in the academic setting is primarily an informal process, many biopharmaceutical 75% of ensuring quality in the basic research and discovery space. These practices range from basic document control to strict process control systems resembling Good % Respondents companies have formalized some aspects 50% No Formal Quality System Have Formal Quality System Laboratory Practices that the FDA mandates in later stages of the drug development 25% process.75 These industry practices can include defined experimental replication procedures or staff training requirements to perform daily research functions. Many academic laboratories and institutions, however, do not have such formal quality systems76 (Figure 12). No Formal Quality System 0% Academia Industry Interview participants (N=30) were asked about whether their laboratory, institution, department, or company had a formalized quality system pertaining to basic and discovery research. “Formal” refers to presence of policies outlined in writing that were in some way enforced. 75. Code of Federal Regulations, Title 21, Part 58. 76. A quality system is a formalized way to ensure that practices that lead to better outcomes are consistently performed. ASQ Glossary. [Internet] [cited 2013 Jul 1]. Available from http://asq.org/glossary/. 24 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Negative Effects of Irreproducibility: The Case for Standards Collaborations can rapidly break down when academic results are not reproducible in “If we had industry-wide standards for laboratory industrial laboratories because, no matter how processes or ensuring that everyone performing promising academic published results might research is actually qualified to perform the research be, irreproducible results cannot be developed they are doing, we could say to laboratories or into commercializable products. Industry stakeholders report that differences in quality institutions ‘You have to follow these guidelines to assurance practices, the quality of reagents, work with us.’” and in the degree of attention paid to the Biopharma Industry Executive reproducibility of processes can present obstacles to translating academic laboratory results into drug products. Irreproducibility Damages Reputations Irreproducible results can significantly damage the reputations of laboratories, institutions, journals, and funders associated with them. Laboratories whose work cannot be reproduced face reputational concerns that can adversely affect their ability to publish or obtain funding in the future. Names of institutions where the irreproducible research was produced can be connected One academic research leader described an example of irreproducible work that was performed in his laboratory. The irreproducibility was due to a weak signal on a gel in a published experiment that was interpreted as positive in his laboratory and as negative by others. Within a year, a competitive lab published a paper with a better protocol showing a strong signal, indefinitely to irreproducibility. For example, a 2013 review article about XMRV59 describes the investigators of the original 2006 study as “researchers at the Cleveland Clinic and the University of California,” as opposed to referring to the PIs by name. Journals likewise face negative effects on their reputations if a significant amount of their published work is shown to be irreproducible. As concerns about and took credit for the discovery. They even included a figure showing that the original results were irreproducible. The academic research leader suffered some embarrassment within his community and believes that his laboratory lost irreproducibility gather popular attention, being associated with a publicized incident might worsen public perception of the funding organization or agency and the overall public opinion of life science research. some credibility as a result of this incident. “I know the results were right, but now I insist that everything we publish is reproduced by someone in the lab who didn’t work on the project before we submit it for publication.” Academic Research Leader “If we find out three years later that some of the work we are funding was predicated on false or inaccurate data, that’s a terrifying, expensive, and dangerous problem.” Disease Foundation Executive THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 25 Life Science Research Standards: Adding Value and Accelerating Progress Further Applications of Standards to Life Science Research Standards can benefit life science research quality and reproducibility by decreasing unrecognized variance between laboratories, and can serve as an effective tool for disseminating and encouraging consensus-based research practices. As illustrated in Figure 13, it is possible to envision standards pertaining to multiple aspects of life science research. Figure 13. Standards Applications in Life Science Research • Establishment of laboratory or institutional systems that ensure research quality • Standardized routine laboratory processes such as equipment calibration • Reference reagents • Reagent validation Quality Systems Reagents Laboratory Maintenance Assays • Consensus assays • Consensus cut-off values • Reference reagents for calibration Life Science Standards • Consensus reporting frameworks for materials, methods, and procedures • Consensus systems for reporting experimental data and results Legend: Reporting Protocols Data Analysis Good Experimental Practices • Consensus best practices for data analytic techniques • Consensus on best practices for common protocols • Consensus best practices for general concepts in experimental design and performance Materials & Document Standards Primarily Document Standards Standards can have broad applications in life science research. Blue indicates areas where both materials and document standards may be applicable; green indicates areas where primarily document standards are applicable. Standards related to reagents, assays, and laboratory maintenance increase quality and reproducibility by decreasing unintentional (unrecognized) variation between laboratories. Standards related to reagents can include both reference reagents and document standards that describe reagent validation techniques. It is not uncommon for laboratories to use what they believe, based on name or vendor description, to be the same antibody or cell line as another laboratory, only to find out later that the particular lot or batch of the reagent had different biological qualities that lead to variability in results. 26 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Life Science Research Standards: Adding Value and Accelerating Progress Materials standards for certain research reagents (e.g., antibodies, cell lines, animal models) can be used by laboratories and vendors to decrease these unexpected differences. Document standards pertaining to reagent validation can outline consensus-based ways of ensuring that a particular reagent performs “There is a lot of variability in reagents such as antibodies as expected and, likewise, decrease unanticipated variations that lead to irreproducibility. between manufacturers and even between lots. It can take Assays and measurements present another area a significant amount of time to validate a new antibody and sometimes we can never get it to work. Standards to reduce this would be very helpful and save time.” Academic Research Leader where unintended variation can be reduced through biological standards. In many life science research areas, there is little consensus about which assays researchers should use. There might not be reagents available to calibrate the assays in order to ensure that results obtained in one laboratory are comparable to another. Furthermore, there might be variability in cut-offs for considering results positive or negative. Document standards can enable consensus around assays and measurements. For example, a standard can represent the scientific community consensus on which assay will be used to measure activity of a particular bacterial toxin and on the cut-offs for positive and negative “There is a need to standardize assays used in [Clostridium] difficile research. Papers use different results. Additionally, the development of techniques for measurement, and this makes it difficult materials standards for assay calibration can to compare and synthesize results across different particularly benefit fields that use novel and publications. I believe a consensus method would emerging technologies, where such standards are frequently not available. increase the pace of research in this area.” Academic Researcher Standards surrounding laboratory maintenance can define expectations for certain routine “housekeeping” practices. These practices can include the accurate documentation of laboratory protocols and procedures, data recording (e.g., maintenance of laboratory notebooks, deciding how long original data is kept), equipment calibration (e.g., annually ensuring that all pipettes in a laboratory are accurate), checking for reagent expiration, and not using expired reagents. Standards can also outline good practices for common laboratory activities, such as cell or microorganism culture and cell line or microorganism authentication. A limited number of standards pertaining to reagents, assays, and laboratory maintenance are already available. Some materials standards, primarily microorganisms, are available to researchers through the World Health Organization (WHO), NIH, and private bioresource centers. Document standards are gaining increasing adoption. One example is the ATCC cell line authentication standard that describes a consensus procedure for unambiguous authentication and identification of human cell lines using a specific type of analysis (short tandem repeat [STR] profiling).77 (See call-out box on next page.) However, many opportunities for achieving community consensus and alignment remain and can serve to reduce unintentional variability and improve reproducibility between laboratories. 77. ANSI/ATCC ASN-0002-2011. 2012. Authentication of Human Cell Lines: Standardization of STR Profiling. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 27 Life Science Research Standards: Adding Value and Accelerating Progress Life Science Standards at Work: Cell Line Authentication Cell lines are cells either from cancer or normal tissues that can proliferate under laboratory conditions. Initial experiments on a particular tumor type (e.g., breast cancer) are usually performed using these cells. A large number of cells lines representing a variety of tumor types exist. Laboratories depend on visual examination of the cells and vial labels to ensure that they are performing each experiment on the correct cell type. However, even for an expert, it can be difficult or impossible to determine with certainty from which type of tumor a particular cell line originates.78 Because cells are repeatedly grown, frozen, and stored by laboratories, cross-contamination and errors can occur, resulting in experiments being performed on an incorrect cell type. Examples of using incorrect cell lines in research have been reported for a variety of tumor types. A database maintained by the International Cell Line Authentication Committee (ICLAC) lists over 350 cell lines that have been misidentified or cross-contaminated and publications based on these cell lines.79 Although many of these have been contaminated by the HeLa cell line, the problem is even more pervasive. For example, several common esophageal cell lines that have served as a basis for over 100 publications and several clinical trials have been shown to be from other parts of the body (e.g., lung and colon).80 In 2011, a novel gastric MALT lymphoma cell line was described. It was considered groundbreaking because no cell lines were available to model this disease. Recently, however, this cell line was shown to actually be a misidentified, well-known cell line of a different lymphoma type.81 Misidentification of cell lines can be prevented by cell line authentication. Authentication is based on determining the genetic signature of a particular cell line and comparing it to established databases to ensure that the cell line used by a laboratory matches the expected signature.82 In 2011 and 2012, an international group of scientists from academia, regulatory agencies, major cell repositories, government agencies, and industry collaborated to develop a standard that describes optimal cell line authentication practices, ANSI/ATCC ASN-0002-2011: Authentication of Human Cell Lines: Standardization of STR Profiling.83 More recently, multiple journals including Nature,84 several American Association for Cancer Research (AACR) journals,85 and the International Journal of Cancer86 now require or strongly recommend cell line authentication. In addition to decreasing inter-laboratory variation in reagents, assays, and daily laboratory operations, standards can also help researchers reach voluntary consensus on general scientific practices and disseminate these optimal practices. For example, standards can outline general good experimental practices, such as a suggested number of replicates, blinded experimentation, appropriate selection of controls, or selection of experimental systems; standards can also serve as a reminder and educational tool about these practices. Standards can also be a vehicle for defining optimal protocols for common 78. 79. 80. 81. 82. 83. 84. 85. Capes-Davis, A. et al. Int J Cancer. 2013; 132(11):2510–2519. Capes-Davis, A. et al. Int J Cancer. 2010; 127(1):1–8. Boonstra, J. et al. J Natl Cancer Inst. 2010; 102(4):271–274. Capes-Davis, A. Genes Chromosomes Cancer. 2013; 52(10):986–988. ANSI/ATCC ASN-0002-2011. 2012. Authentication of Human Cell Lines: Standardization of STR Profiling. Kerrigan, E. “Working Together to Eradicate Cell Line Misidentification and Contamination: New Practices and Policy.” DDNews. 2012 Sep. Nature Editorial Board. Nature. 2013; 496:398. AACR Journals: Instructions for Authors. [Internet] [cited 2013 Aug 5]. Available from http://cancerres.aacrjournals.org/site/misc/ifora.xhtml#celllineuse. 86. International Journal of Cancer Author Guidelines. [Internet] [cited 2013 Aug 5]. Available from http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1097-0215/homepage/ForAuthors.html. 28 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Life Science Research Standards: Adding Value and Accelerating Progress experimental procedures. Additionally, standards can apply to data analysis by outlining consensus practices for the use of particular statistical methods, handling of outliers, and even the need for biostatistician involvement. Standards pertaining to submission of results to peer-reviewed journals (e.g., reporting guidelines) or to separate databases can facilitate consistency and completeness of reporting for experimental methods and results. Although some standards such as Minimal Information About a Microarray Experiment (MIAME)87 already exist in this area, there is tremendous variance between journals and even between reviewers in the same journal about the level of detail reported for many types of experiments.88 Standards could help journals align around the types of data that are necessary to describe the experimental conditions, analysis, and results. Finally, standards can serve to establish systems that promote and ensure consistent quality and the reproducibility of research results. For example, standards can outline quality systems for performing life science research. Standards pertaining to quality systems (such as the ISO [International Organization for Standardization] 9000 standards) generally do not explicitly state what constitutes a high-quality outcome (e.g., that every experiment submitted for publication has an appropriate control). Instead, these standards establish institutional frameworks for checking quality consistently based on criteria that may be defined either by other standards or by each institution or laboratory. For instance, such a standard may state that an institution or a laboratory must have a documented internal process for ensuring that all experiments submitted for publication have appropriate controls. Quality system standards have been successfully and broadly adopted in other industries. For example, over 1 million organizations adhere to quality standards and are ISO-9000 certified,89 and over 7,000 laboratories in the United States and internationally adhere to the College of American Pathology (CAP) standards.90 There has also been some progress toward developing these types of standards for life science research. In particular, the American Society for Quality (ASQ) produced a technical report, ASQ TR1-2012: Best Quality Practices for Biomedical Research in Drug Development, in 2012. That document outlines a quality system aimed at non-regulated biomedical research, and could serve as a foundation for standards in this area. 87. MIAME is a standard for reporting of microarray data developed in 2001. Please see Appendix I for additional information. 88. The variation in implementation and effectiveness reporting guidelines has been particularly well-documented in literature pertaining to clinical trials. See Hirst, A. and Altman, D.G. PLOS ONE. 2012; 7(4):e35621. These findings are also summarized in Tugwell, P.A. et al. J Clin Epidemiol. 2012; 65(3):231–233. 89. ASQ. “25 Years of ISO 9000.” [Internet] [cited 2013 Sep 16]. Available from http://asq.org/blog/2012/03/25-years-of-iso-9000/. 90. CAP. ”About the CAP: Laboratory Accreditation Program.” [Internet] [cited 2013 Sep 16]. Available from http://www.cap.org/apps/cap.portal?_nfpb=true&_pageLabel=accreditation. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 29 Life Science Research Standards: Adding Value and Accelerating Progress Systemic Benefits of Standards Life science research standards can be a unifying driver of quality improvement efforts and have the potential to benefit all stakeholders (Figure 14). Figure 14. Benefits of Standards Academic Labs & Institutions Industry & Investors • Ensure best practices incorporated into research process • Reduce target validation and development failures • Produce more reproducible results • Improve translatability and commercialization of discoveries • Decrease misinformation • Enhance collaboration with academia • Protect reputation • Enhance collaboration with industry Government & Non-Profit Funders • Facilitate grant review process • Improve return on investment • Optimize use of funds Standards = Reproducibility Journals & Professional Societies • Facilitate peer review process • Optimize use of financial resources • Protect reputation • Protect reputation • Decrease misinformation • Improve public perception of life science research • Improve public perception of life science research More Efficient Use of Resources and Time Protection of Reputation More Favorable Public Opinion of Research Standards can benefit the quality of the research process by strengthening current checkpoints and counteracting forces that lead to quality challenges (Figure 15). Document and material standards (e.g., reference reagents) can serve as tools to disseminate best practices to laboratories and institutions, resulting in the overall elevation of minimum experimental quality and facilitate the meaningful comparison of results between different laboratories. Standards can also ease the evaluation of quality and experimental robustness during the grant review and peer review processes by creating a unifying framework, developed and consented to by experts in the field, and available to a broad range of stakeholders. A standards framework can enable stakeholders to ascertain that best practices are being followed without delving into the details of how they are followed. Over time, adoption of optimal, consensus-based practices and materials standards can lead to increasingly reproducible results and lower reliance on time-consuming self-correction. 30 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Life Science Research Standards: Adding Value and Accelerating Progress Figure 15. Standards Counteract Quality Challenges Increasing Quality Challenges Improvement Through Standards Research Findings Research Findings • Decreasing grant application page number ➡ • Easier evaluation of quality and experimental robustness Grant Review • Less resources per paper leads to less robust experimental design ➡ • Better minimum experimental design quality • Decreased bias and variability Intralaboratory Pl Review • Increasing complexity • Inconsistent review metrics ➡ • More effective peer review and ability to judge complex work Peer Review • No incentive for replication (funding or publication) ➡ • Decreased variability • Easier to compare data Colleague Replication Scientific Knowledge • Decreased time due to increasingly translational focus ➡ • Less reliance on time-consuming self-correction Scientific Knowledge Self-Correction Legend: High-Quality Work Low-Quality Work The funnels represent the research process, with circles representing high-quality (yellow) or low-quality (blue) work. The research flows through the checkpoints described on the right and most low-quality work is filtered out. With increasing quality challenges, the filters become more permeable and allow more low-quality work to pass through. Standards, represented by orange arrows, can counteract the quality challenges and strengthen the current quality checkpoints (orange), resulting in less low-quality research. By increasing reproducibility, standards can make research results more translatable and facilitate collaboration between industry and academic laboratories. Industry stakeholders report that they would prefer to collaborate with academic laboratories that follow standardized processes and use reliable reagents because this would align these laboratories closer to internal quality processes routinely practiced by industry (described above). Working with laboratories that adopt best practices through consensus standards will increase the reproducibility of their research and reduce technology and process transfer failures. Collaborating with these laboratories could also help to decrease uncertainty of outcomes inherent in academic-industry research partnerships. Finally, standards can unify and accelerate current and future efforts to improve life science research quality. Stakeholders have started to address irreproducibility in multiple, frequently disjointed ways, including action by journals, funding agencies, and independent organizations.91 For example, Nature recently launched a Reproducibility Initiative,92 and the NIH has held several workshops to address the 91. Please see Appendix I for detailed descriptions of representative efforts to enhance life science research quality and reproducibility. 92. Nature Editorial Board. Nature. 2013; 496:398. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 31 Life Science Research Standards: Adding Value and Accelerating Progress issue of irreproducibility.93 The 2013 Seventh Peer Review Congress had several Figure 16. Standards are a Unifying Driver of Quality Improvement Efforts presentations and posters dedicated to this and related topics.94 Stakeholders interviewed for this Report can also envision multiple other ways of improving reproducibility such as oved Impr tion a docs in postEduc improved education. Current efforts to improve research quality and reproducibility in the life sciences are laudable and have the potential to be highly Impro v Sharin ed g of D a ta “C Institutional Policies g ers catin arch “Edu ior rese f what is n o uce or ju ncepts prod o the c essary to le work nec ducib long ro p a re ld go wou lder keho .” way ustry Sta Ind effective. However, these efforts are highly “Institutions could implement programs to improve quality and advertise them to potential industry partners.” onsis report tent ing c used ancer sub of would ty data make pes a lot the ea comp are a sier to nd sy nthes Disea ise.” se Fo “Com unda tion mon are n d ecess ata form ats a datas ets.” ry to share A Quality Expert Journal Review Policies cadem “Publication is the goal of academic research, so anything impacting that will be powerful.” ic Re searc h L ea der Academic Research Leader dependent on the activities of individual journals, institutions, or organizations, and place a significant burden on these stakeholders. Because these efforts are disconnected, each stakeholder must invest licies d er Po de Fund NIH demanld do u e “If th eople wo of the d p ch this, . So mu controlle it g is ould in d ere w .” fun th , e em tiv by th r alterna h Leader rc othe esea be no ademic R Ac anew in identifying best practices and developing its own definitions, guidelines, or Quality Sy stem Greater Experimental Robustness “It is a matter of applying the basic principles of the scientific method and doing it consistently, like using statistically valid sample size and appropriate data analysis techniques.” s “A formal qu is the best ality system way of reducing uncontrolle d variables and insur ing appropria te re keeping, eq corduipment calibratio n and training.” Standards Expert Industry Stakeholder systems. By providing a common framework of practices, biological materials, and methods developed by or consented to by key stakeholders in a particular area, standards can facilitate existing efforts to improve life science research quality and reproducibility. Standards “Over time, science self-corrects. But if you have standards to improve quality of laboratory procedures, to improve basic techniques that result in inconsistencies, this process does not always need to happen.” Standards Expert Greater Experimental Robustness • Raise awareness of proven methodologies and lower uncontrolled variation Journal Review Policies • Harmonize and simplify review policies Funder Policies Overall, standards are broadly applicable across life science research and can significantly benefit the entire field. Standards can help the life science research community align around consensus-based best practices, reduce uncontrolled variance, and improve Quality Systems Improved Sharing of Data • Enable common reporting frameworks and formats • Simplify quality assessment during grant review process • Enable implementation and accreditation process Improved Education Institutional Policies • Provide consistent framework for policies • Serve as an educational tool and ensure eduction reproducibility, resulting in more efficient use of resources, less misinformation, and a more favorable public opinion of research. 93. Landis, S.C. et al. Nature. 2012; 490(7419):187–191. 94. Peer Review Congress. [Internet] [cited 2013 Sep 15]. Available from http://www.peerreviewcongress.org/2013/Final-Program.pdf. 32 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Moving Forward to Solve Irreproducibility While life science research has made tremendous advances over time, traditional methods for ensuring quality are coming under increasing pressure as the field progresses and the funding environment worsens. Irreproducibility of life science research findings is a pervasive, systemic, and expensive problem that affects those directly involved in research and publication, as well as funders and the general public. Key stakeholders—academic laboratories and institutions, industry and industry associations, investors, government agencies, charitable foundations, journals and professional societies, standards development organizations, and policy and patient advocacy organizations— associate irreproducibility with poor quality, and are voicing significant concerns about the effects of irreproducibility on their work, on the use of public and private resources, and on public perception of life science research. Thus far, however, there has been a chasm between the pervasiveness and seriousness of the problem and efforts to find solutions. To date, there have been no coordinated endeavors with a national or international scope to engage all stakeholder groups in order to find effective ways to increase reproducibility across the life sciences research landscape. Emerging efforts, such as the Nature Reproducibility Initiative, are important and will surely benefit the field. Nevertheless, the prevalence and magnitude of this problem requires a systemic, unifying effort. Life science standards can provide the basis for a collaborative, global movement to improve reproducibility. • Standards are an essential platform that can elevate the quality of the field by encouraging stakeholders to identify and disseminate best practices and optimal materials. • Standards can serve as a unifying driver for current and future efforts to improve reproducibility by providing stakeholders with a common framework of expert, consensus-based opinions. Uniting the diverse life science research community in a standardization effort will be challenging. Nonetheless, multiple other fields have successfully incorporated standards to improve quality by associating with core organizations. For example, over a million organizations are ISO 9000-certified and thousands of laboratories adhere to the College of American Pathology standards. The development and implementation of standards, whether as adherence to written documents or use of reference materials, requires community consensus and alignment around both the necessity for standards and their content. Ultimately, expanded adoption of life science standards will require: • Educational initiatives to raise stakeholder awareness of the purpose and benefits of biological standards and understanding of the standards development process; • Opportunities and forums for stakeholders to identify areas in the life sciences where accelerated standards adoption could provide maximum benefit; • Engagement of stakeholders with standards development organizations or material reference providers in the development of specific standards; and • Development of effective policies and practices within the life science research community to ensure the proactive development and periodic updating of biological standards. The life science research community is creative, inventive, and innovative. It is a community that has found cures for diseases and extended human life. This community can solve the problem of irreproducibility, and a solid standards framework can unify and accelerate this global effort. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 33 Appendix I: Supporting Materials Description of Current Academic Life Science Research Process Government research funders (e.g., NIH) and non-government funders, such as disease or other charitable foundations, all use the peer review95 process to evaluate grant proposals. Multiple criteria, such as the potential importance of work to the field and adherence to funders’ guidelines, are used. This review typically forms the basis for deciding whether the funding will be granted. After funding for a particular research project is obtained, the PI is responsible for controlling the quality of the research performed and of any papers submitted for publication. He or she must balance ensuring research quality with allowing sufficient freedom to promote innovation. The PI must also maintain a suitable educational environment because academic research laboratories are primarily staffed by graduate and post-graduate students.96 After a manuscript is prepared, it is submitted to a journal for publication. Checkpoints here are particularly important in ensuring the quality of research because the researchers’ publication records, including the number of publications and the reputation of the journal where the work is published, are commonly used to determine whether a researcher is hired, promoted, or receives tenure.97 These considerations may also be taken into account during the grant-funding process. Journals use a combination of peer and editorial review to determine which papers will be accepted for publication. Journals also take the novelty and importance of results to the field into account. Extensive variability exists in requirements (e.g., reporting guidelines) for publication as well as criteria used by different journals and reviewers to evaluate work. Publications release and disseminate results into the broader scientific community where findings and interpretation face additional scrutiny. Feedback from the life science community suggests that researchers believe that when the journal peer review process fails and erroneous data or interpretations get published, highly important or controversial results will be either replicated and confirmed or refuted by other laboratories within a few years. If the findings are reproducible, other laboratories will often use them as the basis for additional research, resulting in further publications and the growth of the field. If findings are not reproducible, the situation becomes more variable. When the original hypothesis is of tremendous significance, the irreproducibility of results is usually published, although this might take several years. For results of less perceived significance, the inability to reproduce findings is frequently not discovered nor reported. Over the longer term, work that builds upon the conclusions of published research either confirms or refutes them, resulting in a process of scientific “self-correction.” This process may take many years, is not infallible, but acts to eliminate erroneous hypotheses and results from the scientific worldview.98,99 Academic researchers are typically comfortable with the concept and process of self-correction because it stems from the basic trial-and-error approach inherent in the application of the scientific method. Document Standards The International Organization for Standardization (ISO) and American National Standards Institute (ANSI) define a standard as “a document, established by consensus and approved by a recognized body, that provides, for common and repeated use, rules, guidelines or characteristics for activities or their results, aimed at the achievement of the optimum degree of order in a given context.”100 95. Peer review is a process in which the work of one scientist is evaluated by other scientists with expertise in the field, generally recruited on a voluntary basis. 96. Rohn, J. Nature. 2011; 471:7. 97. Mann, S. AAMC Reporter. 2013 Jan. 98. Ioannidis, J.P. Persp Psych Sci. 2012; 7(6):645–654. 99. The Economist. [Internet] [cited 2013 Oct 21]. Available from http://www.economist.com/news/briefing/21588057-scientists-think-science-self-correcting-alarming-degree-it-not-trouble. 100. ISO/IEC Guide 2:2004, definition 3.2. 34 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Appendix I: Supporting Materials “Approval by a recognized body” is the main factor that separates voluntary document standards from consensus guidelines or best practices.101 Guidelines, recommendations, or best practices can be developed by any group and do not necessarily need to engage all stakeholders that the guideline may affect. In contrast, standards are commonly developed by standards development organizations (SDOs) or consortia. SDOs are collaborative organizations that meet international requirements for openness, balance, consensus, and due process, and provide all affected stakeholders with an opportunity to participate in the development of standards.101 These requirements are outlined in a system of standards pertaining to SDOs, and SDOs are accredited to follow these standards by national “Standards should be based on the consolidated results of science, technology and experience, and aimed at the promotion of optimum community benefits.” International Organization for Standardization in ISO/IEC Guide 2:2004, definition 3.2 and international standards bodies. ANSI is both the U.S. national standards body and representative to ISO, which is the worldwide federation of national standards bodies. ISO, which was established in 1947 and is based in Geneva, Switzerland, draws members from over 145 countries. There are currently over 200 SDOs in the United States participating in this system, which helps ensure that all major perspectives in a particular field are elicited during the standard development process, and that all the major stakeholders are engaged. Consortia are generally not accredited, but follow the spirit of this system by engaging all key viewpoints in a particular field during standards development. Document standards have been adopted across multiple industries. For example, the National Electrical Manufacturers Association (NEMA) publishes standards for electrical plugs and sockets in the United States to ensure that all electrical devices intended for home use have similar plugs that are compatible with commonly installed sockets.102 Standards describing systems that companies need to put in place in order to produce high-quality products and services, such as ISO 9000 series or ASQ Quality Standards, are broadly used in most industries to improve and maintain performance. College of American Pathology (CAP) Standards outline processes and procedures necessary to maintain quality in clinical diagnostic testing and are used by clinical laboratories throughout the United States. Accreditation is one way to encourage compliance with voluntary standards. For example, a company may choose to work only with suppliers that have ISO 9000- or ASQ-accredited quality practices because this accreditation increases the likelihood that the supplier will deliver a high-quality product. As a result of voluntary inducement and community expectations, over one million companies are ISO 9000 series-certified.103 The Centers for Medicare and Medicaid (a government agency) uses accreditation by Joint Commission or CAP (both of which are independent, non-profit organizations) as a basis for determining whether Medicare will reimburse services from a particular clinical laboratory,104 resulting in almost universal laboratory certification. Materials Standards Materials standards are highly characterized physical substances, such as chemical or biological reagents (known as reference materials or reference reagents), and are routinely used in several areas of life science research and healthcare. Examples of materials standards include reagents for biologics and pharmaceutical development and production, such as well-characterized virus strains to be used for vaccine production (e.g., annual flu vaccines) or in genetic therapy development. Materials 101. 102. 103. 104. ANSI. [Internet] [cited 2013 Sep 17]. Available from http://www.ansi.org/about_ansi/faqs/faqs.aspx. NEMA. NEMA Standards Publication. ANSI/NEMA WD 6-2002 (R2008). ISO. “The ISO Survey of Management System Standard Certifications – 2011.” CAP. About the CAP Laboratory Accreditation Program. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 35 Appendix I: Supporting Materials standards can also be used to test assays105 in order to ensure that assays provide accurate measurements and work appropriately. For example, when a sample is tested for a specific infectious organism (e.g., a bacterial pathogen) in a clinical laboratory, infectious organism reference materials (or derivative commercial reagents) can be used to ensure that the assay is sufficiently sensitive and specific to detect the microorganism. Reference materials can be produced and distributed by several sources, depending on the type of reference material. Sources may include the WHO; government agencies such as the NIH and the National Institute of Standards and Technology (NIST); non-profit entities, including biological resource centers like ATCC; and commercial companies. The exact source varies by the specific material. Standards Example: HbA1c Testing Materials and systems standards are effective in reducing inter-laboratory variability and irreproducibility, as exemplified by the standardization of testing for glycated hemoglobin. Glycated hemoglobin is a marker of how effectively blood sugar is controlled in diabetic patients. Although a direct blood glucose measurement provides information about the patient’s blood sugar control at that moment, glycated hemoglobin measurements provide information for how blood sugars have been controlled during the past month or more. Hemoglobin A1c (HbA1c) is one type of glycated hemoglobin test. The importance of tightly controlling blood sugar in diabetic patients and of HbA1c as a marker of control became broadly accepted as a result of two large clinical trials.106,107 Both trials determined HbA1c values that were associated with a lower risk of diabetic complications, and resulted in recommendations that patients’ blood glucose should be controlled to a degree that would maintain the patient’s HbA1c below these target levels. Figure A. Standardization of HbA1c Testing 8.0 7.5 7.0 % Hb A 6.5 6.0 5.5 5.0 4.5 4.0 3.5 Legend: 3.0 HbA1c HbA1 Total GHb DCCT Target 2.5 1993 1999 2004 2010 Standardization Figure reproduced from Little (2011). Data from College of American Pathology surveys 1993–2010. Each circle, square, or rhombus represents the mean value for testing a control material obtained through distinct method of glycated hemoglobin determination used among survey participants. Bars show 2 standard deviations from the mean. 105. An assay is a biological, biochemical, or chemical way of measuring if and how much of a particular substance, biological molecule, or organism is present. Assays are frequently used in research, public health, or as diagnostic tests. 106. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993; 329:977–986. 107. U.K. Prospective Diabetes Study (UKPDS) Group. Lancet. 1998; 352:837–853. 36 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Appendix I: Supporting Materials At the time these trials came out, laboratories used multiple methods for measuring glycated hemoglobin that produced results that were not directly comparable.108 For example, in 1993, only 50% of laboratories were actually reporting results as HbA1c; the remaining were reporting different measurements such as total glycated hemoglobin. Furthermore, even laboratories using the same methods had substantial variability in results. In essence, the results were irreproducible from laboratory to laboratory. This made it difficult or impossible for clinicians to compare the HbA1c levels obtained from their patients to the target levels from the trials or even to compare the results for a single patient over time if they were obtained in different laboratories (Figure A). For example, if the same patient’s blood sample was sent to two different laboratories, it would not be unusual “HbA1c testing standardization is an example of a success story. The community aligned behind this effort. The clinical societies, laboratory societies, manufacturers, and laboratories themselves – everyone came to understand that the standardization was necessary to provide evidence-based patient care.” Standards Expert for one laboratory to produce results that were below the target level established in the clinical trials, indicating that the patient’s treatment was working appropriately, and for a different laboratory to produce results significantly above the target level, indicating that the treatment was not working and should be increased or changed. Due to the extent of this problem, two professional societies, the American Association for Clinical Chemistry (AACC) and International Federation of Clinical Chemistry (IFCC), created committees to enable the standardization of testing methodologies. These groups created two complementary systems that together serve to standardize manufactured HbA1c assays. The IFCC created reference materials for HbA1c measurement that can be used across different methodologies for calibration, and to ensure that findings obtained though all methodologies are comparable. Manufacturers can obtain calibrators and controls with values assigned by the IFCC and use them to determine how their assay methodology compares to the IFCC reference method. The AACC effort developed a standard protocol109 and created the National Glycohemoglobin Standardization Program (NGSP), a laboratory-based certification program where manufacturers of HBA1c tests can exchange samples with special NGSP laboratories in order to certify that the results of the test are consistent with the results obtained using the standard protocol. Thus, the IFCC program ensures that each manufacturer can trace its assay to an accurate base through a reagent standard-based system, whereas the NGSP program certifies that the assay accuracy is within an acceptable range of results obtained using a standard protocol. Together, these standardization systems significantly improved the accuracy, reproducibility, and clinical value of HbA1c testing. As Figure A shows, in 2010 with over 3,000 laboratories participating in the CAP survey for HbA1c testing, the variability between different testing methods and within each testing method was significantly lower than in the years before standardization became common (1993 and 1999). Today, clinicians can routinely use and meaningfully compare HbA1c values from different laboratories and be assured that the values they are using to treat their patients are accurately connected to clinical trial results. 108. Little, R. Clin Chem. 2011; 57(2):205–214. 109. Little, R. Clin Chem. 2001; 47(11):1985–1992. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 37 Appendix I: Supporting Materials Standards Example: Bacterial Testing of Platelets The development and adoption of standards surrounding the bacterial testing of platelets is an example of how the quality of laboratory practices can be improved by document (written) standards. Platelets are a component of blood donations that are given to patients with low platelet counts in order to prevent bleeding complications. Over 2 million units of platelets were administered to patients in the United States in 2011.110 Because of their biological properties, platelets are stored at room temperature, which encourages the growth of bacteria. If bacterially-contaminated platelets are given to a patient, the patient might get ill and develop a systemic infection or sepsis. Prior to 2004, some blood banks used technologies to detect bacterial contamination in platelets, believing it to be necessary for patient protection, but many blood banks did not. As a result, bacterial contamination of platelet products was acknowledged to be the most frequent infectious risk from transfusion in the United States, and was considered the second most common cause of death resulting from transfusion.111 In response to increasing research pointing to bacterial contamination of platelets as a significant cause of transfusion-related illness and death, in 2004 the AABB, a voluntary standards and accreditation body, developed a standard for limiting and detecting bacterial contamination of platelets. The development of that standard involved extensive input and consensus from the blood banking community. As a result of adoption of this standard by AABB- “At the time when the standard was implemented, accredited blood banks, septic transfusion some people were resistant because they felt it reactions per 100,000 transfusions had decreased increased blood bank costs and disrupted operations. by 70%. There was also a similar decrease in deaths from transfusion-related sepsis.112 Overall, Of course it wasn’t perfect, but looking back now, the AABB standards surrounding bacterial testing having the standard is a really good thing. It improved of platelets resulted in better care and saved lives transfusion safety and cut the risk associated with of platelet recipients. bacterial contamination of platelets more than in half.” Blood Banking Expert Standards Example: Laboratory Animal Welfare The most well-developed standards and accreditation system exists to protect laboratory animal welfare. In the United States, this system is centered around the Guide for the Care and Use of Laboratory Animals, a collection of detailed standards for appropriate laboratory animal care initially published in 1963 and revised multiple times thereafter. The latest revision (the 8th edition) was published in 2011 by the Committee for the Update of the Guide for the Care and Use of Laboratory Animals,113 which was appointed in 2008 by the National Research Council.114 The Committee included research scientists, veterinarians, and nonscientists representing biomedical ethics and the public’s interest in animal welfare. Funding for development and publication of the Guide was received from government agencies and non-profit organizations, as well as from the pharmaceutical industry. Adherence to the principles in the guide is required for institutions that receive Public Health Service support (including grants from NIH and CDC), and is monitored internally by each institution through self-regulating Institutional Animal Care and Use Committees.115 Additionally, an independent voluntary 110. 111. 112. 113. U.S. Department of Health and Human Services. The 2011 National Blood Collection and Utilization Survey Report. Hillyer, C.D. Hematology Am Soc Hematol Educ Program. 2003:575–589. AABB. Public Conference – Secondary Bacterial Screening of Platelet Components, 7/17/12. Committee for the Update of the Guide for the Care and Use of Laboratory Animals. Guide for the Care and Use of Laboratory Animals, 8th Ed., NRC 2011. 114. The National Research Council (NRC) is a group of non-profit institutions that include the National Academy of Sciences and the Institute of Medicine. 115. IACUC. [Internet] [cited 2013 Aug 18]. Available from http://www.iacuc.org/aboutus.htm/. 38 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Appendix I: Supporting Materials accreditation system administered by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC International) exists to ensure that institutional self-monitoring is optimal. The AAALAC accreditation also requires adherence to two other standards when applicable: Guide for the Care and Use of Agricultural Animals in Research and Teaching (Ag Guide)116 and the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (ETS 123).117 Standards Example: Data Reporting Frameworks Voluntary standards that have been recently adopted in life science research are focused on establishing “Before the MIAME criteria, we found that even for the data-reporting frameworks. Examples of these include same experimental dataset, the analytic results could only the Minimal Information About a Microarray Experiment be reproduced approximately 10 to 30% of the time (MIAME, developed in 2001),118 Minimum Information because the information in the methods section was not sufficient to ensure that two investigators were really performing the analysis similarly. This is why our journal adopted the standard as a condition for publication.” Journal Editor about a MARker gene Sequence (MIMARKS) and Minimum Information about Any (x) Sequence (MIxS, developed in 2011),119 Minimal Information About a Proteomics Experiment (MIAPE, developed in 2007),120 and Minimum Information about a Flow Cytometry Experiment (MIFlowCyt, developed in 2008).121 Multiple other reporting guidelines for a broad range of experimental techniques are also available.122 These standards, developed by consortia rather than SDOs, apply to high-throughput technologies that generate large datasets of results that are then analyzed using complex bioinformatic techniques. Multiple sources of variability exist in experimental techniques used to generate the data and in assumptions that can be made during the data analysis process (e.g., where to set the cut-off for how much signal intensity is necessary to consider a result positive). These standards encourage more detailed reporting with a particular focus on steps that are common sources of inter-laboratory variability and, as a result, these standards improve reproducibility between laboratories. The MIAME standard has the broadest adoption, and adherence to this reporting framework is currently required by over 50 journals, including high-profile publications such as Cancer Research, Cell, The Lancet, and Nature.123 There are multiple ongoing standards efforts in this area, including Core Information for Metabolomics Reporting (CIMR), a standard to specify minimal guidelines for reporting metabolomics data that is being developed by the Metabolomics Standards Initiative (MSI),124 and the Standard Reference System for Information on Bioinformatics Data Structures that is being developed by IEEE (Institute of Electrical and Electronics Engineers) to help support exchange and comparison of biological data sets in the life sciences.125 A recently-formed international initiative, the Global Alliance to Enable Responsible Sharing of Genomic and Clinical Data, intends to develop technical and other standards to enable sharing of human genome sequencing data for research purposes in a “secure, controlled and interpretable manner.”126 116, 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. Federation of Animal Science Societies (FASS). [Internet] [cited 2013 Oct 22]. Available from http://www.fass.org/docs/agguide3rd/Ag_Guide_3rd _ed.pdf. European Treaty Series (ETS). [Internet] [cited 2013 Oct 22]. Available from http://conventions.coe.int/treaty/en/treaties/html/123.htm. Brazma, A. Nature Genetics. 2001; 29:365–371. Yilmaz, P. Nature Biotechnol. 2011; 29:415–420. Taylor, C. Nature Biotechnol. 2007; 25:887–893. International Society for Advancement of Cytometry (ISAC). Minimum Information about a Flow Cytometry Experiment: MIFlowCyt 1.0. Minimum Information for Biological and Biomedical Investigations (MIBBI). [Internet] [cited 2013 Aug 18]. Available from http://mibbi.sourceforge.net/portal.shtml. Functional Genomics Data Society (FGED). [Internet]. [Cited 2013 Aug 18]. Available from http://www.mged.org/Workgroups/MIAME/journals.html. Metabolomics Society. [Internet] [cited 2013 Aug 18]. Available from http://msi-workgroups.sourceforge.net/. IEEE. [Internet] [cited 2013 Aug 18]. Available from http://standards.ieee.org/develop/wg/1953_WG.html. Global Alliance to Enable Responsible Sharing of Genomic and Clinical Data. Creating a Global Alliance to Enable Responsible Sharing of Genomic and Clinical Data. White Paper, 2013 Jun 3. THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 39 Appendix I: Supporting Materials Initial Efforts to Increase Reproducibility Journal publication requirements, particularly those established by prestigious and respected journals, are a powerful incentive for scientists. Recently, several journals have made efforts to combat irreproducibility of published findings, primarily by encouraging more detailed and complete reporting of methods and procedures. The Nature Reproducibility Initiative,127 for example, centers on an 18-step checklist that includes steps to ensure that authors disclose sufficient technical and statistical information. It is particularly focused on experimental and analytical design elements that the editors consider “crucial for the interpretation of research results but are often reported incompletely.” In conjunction with this initiative, Nature and several associated journals also plan to abolish space restrictions on the methods section in each paper. The NIH has held several recent workshops to address issues of reproducibility. In 2012, NIH’s National Institute of Neurological Disorders and Stroke (NINDS) convened major stakeholders to discuss improving the methodological reporting of animal studies in grant applications and publications.128 The workshop recommended “that at a minimum studies should report on sample-size estimation, whether and how animals were randomized, whether investigators were blind to the treatment, and the handling of data.” A National Cancer Institute (NCI) Board of Scientific Advisors meeting in March 2013 included two presentations focused on reproducibility, including “Data Replication: How Reliable Are the Published Results of NIH-Funded Research”129 by Dr. Harold Varmus, the Director of NCI. Further activity in this area is expected as senior NIH officials assess input from the agency’s 27 institutes before conferring with NIH director Francis Collins.130 The NIH is also considering a variety of proposals, and to date, consensus on the best course of action does not appear to have been reached. Although NIH efforts to consider reproducibility are in the early stages, they may have significant impact on the life science research community in the future due to the potential connection of these initiatives to grant funding. Independent organizations have also started efforts to improve research reproducibility. The Reproducibility Initiative131 is a collaborative effort of several organizations including the Science Exchange, a website that is dedicated to connecting scientists to services they may need; PLOS ONE, a major open-source journal; Mendeley, a reference manager; and FigShare, a data storage and sharing service. To participate, investigators can submit their publications to the website and pay for the independent replication of their research. The Reproducibility Initiative has collected nearly 2,000 author consents to reproduce their work, and recently announced that it had received $1.3 million from the Laura and John Arnold Foundation to assess 50 of the highest-impact cancer findings published between 2010 and 2012.132 The U.K.-based EQUATOR (Enhancing the QUAlity and Transparency Of health Research) Network, formed in 2008, “seeks to improve the reliability and value of published health research literature by promoting transparent and accurate reporting and wider use of robust reporting guidelines.”133 The Network is directed by an international Steering Group that brings together researchers, medical journal editors, peer reviewers, developers of reporting guidelines for health research, research funding bodies, and other collaborators concerned with improving the quality of research publications and of the research process itself. Although EQUATOR is primarily focused on reporting of research involving human subjects, the organization’s scope also includes guidelines related to pre-clinical animal studies.134 127. 128. 129. 130. 131. 132. Nature Editorial Board. Nature. 2013; 496:398. Landis, S.C. et al. Nature. 2012; 490(7419):187–191. [Internet] [cited 2013 Aug 8]. Available from http://deainfo.nci.nih.gov/advisory/bsa/bsa0313/. Waldman, M. Nature. 2013; 500:14–16. [Internet] [cited 2013 Aug 14]. Available from https://www.scienceexchange.com/reproducibility/. [Internet] [cited 2013 Oct 21]. Available from http://blog.scienceexchange.com/2013/10/ reproducibility-initiative-receives-1-3m-grant-to-validate-50-landmark-cancer-studies/ [cited 2013 Oct 21]. 133. [Internet] [cited 2013 Oct 8] Available from http://www.equator-network.org/. 134. [Internet] [cited 2013 Oct 9] Available from http://www.equator-network.org/?post_type=eq_guidelines&eq_guidelines_ study_design=animal-pre-clinical-research&eq_guidelines_clinical_specialty=0&eq_guidelines_report_section=0&s=. 40 THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH Appendix II: List of Acronyms AAALAC AABB AACC AACR ACOG ADDF ANSI ASQ ATCC CAP CDC CF CFS CIMR EQUATOR ETS FASEB FASS FDA GBSI HbA1c HIV IACUC ICLAC IEC IEEE IFCC IND ISAC ISO MIAME MIAPE MIBBI MIFlowCyt MIMARKS MIxS MSI NCI NEMA NINDS NGO NGSP NIH NIST NRC NSF PI RePORT SDO STR UKPDS WHO XMRV Association for Assessment and Accreditation of Laboratory Animal Care (previously known as American Association of Blood Banks) American Association for Clinical Chemistry American Association for Cancer Research American Congress of Obstetricians and Gynecologists Alzheimer's Drug Discovery Foundation American National Standards Institute American Society for Quality American Type Culture Collection College of American Pathology Centers for Disease Control and Prevention cystic fibrosis chronic fatigue syndrome Core Information for Metabolomics Reporting Enhancing the QUAlity and Transparency Of health Research Network European Treaty Series Federation of American Societies for Experimental Biology Federation of Animal Science Societies Food and Drug Administration Global Biological Standards Institute hemoglobin A1c human immunodeficiency virus Institutional Animal Care and Use Committee International Cell Line Authentication Committee International Electrotechnical Commission Institute of Electrical and Electronics Engineers International Federation of Clinical Chemistry Investigational New Drug International Society for Advancement of Cytometry International Organization for Standardization Minimal Information About a Microarray Experiment Minimal Information About a Proteomics Experiment Minimum Information for Biological and Biomedical Investigations Minimum Information about a Flow Cytometry Experiment Minimal Information About a MARker gene Sequence Minimum Information about Any (x) Sequence Metabolomics Standards Initiative National Cancer Institute National Electrical Manufacturers Association National Institute of Neurological Disorders and Stroke non-governmental organization National Glycohemoglobin Standardization Program National Institutes of Health National Institute of Standards and Technology National Research Council National Science Foundation principal investigator Research Portfolio Online Reporting Tool standards development organization short tandem repeat U.K. Prospective Diabetes Study Group World Health Organization xenotropic murine leukaemia virus-related virus THE CASE FOR STANDARDS IN LIFE SCIENCE RESEARCH 41 The Case for Standards in Life Science Research: Seizing Opportunities at a Time of Critical Need The GBSI trademark and trade name is a trademark of Global Biological Standards Institute unless indicated otherwise. This study was commissioned by GBSI. The research and the writing of the Report was completed by Health Advances, a global healthcare strategy firm, in collaboration with Feinstein Kean Healthcare, a strategy and communications firm for life sciences and healthcare. For more information, please contact us at www.gbsi.org [email protected] www.gbsi.org/blog LinkedIn: GBSI Follow us @GBSIorg Our offices are located at: 1020 19th St., NW, Suite 550 Washington, DC 20036 USA + 1 (202) 772-0133
© Copyright 2025 Paperzz