Henry I. Miller

Risk-based oversight of experiments in the environment

THHE NEW BIOLOGY HAS COME OF AGE. BASIC RESEARCH IN fields ranging from immunology to plant biology has been transformed so as to be almost unrecognizable to those whose biology education ended before 1970. The spillover into commercial development likewise has been remarkable. Hardly a week passes without news of some new advance in an area such as therapeutics, vaccines, or plants and animals for food, feed, or fiber. These uses of biotechnology in contained laboratories, pilot plants, greenhouses, and production facilities have engendered little controversy. The National Institutes of Health Guidelines for Research Involving Recombinant DNA have exempted from oversight more than 95% oflaboratory experiments (1); this has allowed organisms of low risk to be handled under modest containment conditions that permit large numbers of living organisms to be present in the workplace and even to be released from the laboratory (2). Despite extensive work in thousands of laboratories throughout the United States with millions of individual genetic clones, there has been no report of their causing a human illness nor any injury to the environment. A bleak spot in this picture is tests in the environment, often termed field trials, planned introductions, or deliberate releases. A number have been subjected to extreme regulatory scrutiny and lengthy delays solely because recombinant DNA techniques were employed in the manipulation of the organism. This has occurred even when the genetic change was completely characterized, benign, and the organism demonstrably innocuous. The ripple effects have been substantial. Investigators have shied away from areas of research that require field trials of recombinant organisms (3); companies have felt compelled to eschew the newest, most precise and powerful techniques in favor of cruder but less regulated ones (4); and investors have avoided companies whose recombinant DNA-derived products require field trials (5). Government agencies have variously regulated new biotechnology products with previously existing regimes or crafted new ones. Whether new or old, certain cardinal principles apply. First, triggers

Regulation of Biotechnology

tee, I have extensive experience in assembling criteria that are used to accord recognition to scientists at or near the Nobel Prize level. We use the number of citations listed for both pertinent people and papers as provided by the Science Citation Index as one of a large number ofcriteria on which to base our decisions. I must believe that responsible, intelligent people in comparable positions who are assessing research performance by university departments would act in a similar fashion, and I only hope that the out-of-hand rejections described in the aforementioned article are not really the position of the people involved but repre-

FDA on transgenic animals--a dog's breakfast?

To the editor: On January 15, the US Food and Drug Administration (FDA) finally published its risk assessment regarding the safety of milk and meat products from cloned animals and their offspring (http://www. fda.gov/cvm/cloning.htm). The situation is not as rosy, however, when it comes to transgenic animals. After 20 years of dithering, the agency has not yet managed to publish a policy statement concerning animals containing a gene from another organism introduced by recombinant DNA techniques. But Larisa Rudenko, a senior official in the FDA’s Center for Veterinary Medicine (CVM), has given a strong hint of the agency’s preferred approach. It is not good news. Rudenko said at Bio 2007 in Boston, MA, that every new genetic construction in an animal that employs gene-splicing technology would require approval for use in the food supply, and that the applicable procedures and regulations would be the same as for drugs used to treat animal diseases. But the introduction of a gene is not the same as the administration of a drug. Not for the first time, FDA is trying to force a square peg into a round hole. Moreover, the CVM’s approach represents a major shift in FDA’s regulation of biotech that will be hugely expensive to animal breeders and detrimental to consumers. When I discussed this in Washington, DC, with John J. Cohrssen, who worked on FDA reform during the 1990s as majority counsel of the US House Commerce Committee, he characterized FDA’s new approach as “complex, arbitrary and dilatory.” Until now, FDA has not regulated farm animals or even animals used for what might be termed ‘medical purposes’. For example, if German shepherds or golden retrievers were bred to enhance traits that made them better seeing-eye or companion dogs, the FDA would not regulate them under its medical device regulations. Nor would a leaner line of pigs be regulated differently from others under the FDA’s food regulations, unless some safety issue were raised. Likewise, for transgenic animals used in medical research, the FDA has not asserted jurisdiction over the hundreds of transgenic rodent lines that are available. The most apposite models for genespliced, or transgenic, animals are the agency’s oversight of traditional foods and food additives; and the production of livestock clones, or identical twins, which FDA confirmed in January were safe to eat. The only transgenic animal currently marketed to the public at large is the ‘Glofish’, a small, tropical, ornamental (aquarium) zebrafish that glows because of the insertion and expression of a gene (from another marine organism, the sea anemone) that synthesizes a beautifully colored fluorescent protein (http://www.glofish.com/)1. The FDA opted not to regulate this organism according to the following rationale: “Because tropical aquarium fish are not used for food purposes, they pose no threat to the food supply. There is no evidence that these genetically engineered zebra danio fish pose any more threat to the environment than their unmodified counterparts, which have long been widely sold in the United States. In the absence of a clear risk to the public health, the FDA finds no reason to regulate these particular fish.” (It is noteworthy that in spite of the fact that Glofish are not eaten and would not survive outside an aquarium, they have been effectively banned by state regulators in California; http://www.fda.gov/ bbs/topics/NEWS/2003/NEW00994.html.) Especially if the standard for becoming subject to regulation is “a clear risk to the public health,” that statement from the FDA would seem to weaken the argument for treating all transgenic animals used for food as though they were being treated with a new drug. A company called Aqua Bounty Technologies (Waltham, MA, USA) has been trying for more than a decade to get FDA approval to market an Atlantic salmon that contains a newly introduced Chinook salmon growth hormone gene engineered to keep it turned on all year round (instead of during only the warmer months, as in nature). This cuts the time to marketable adult weight from 30 to 18 months. The extra gene confers no detectable differences in the salmon’s appearance, taste or nutritional value; it just grows faster. In spite of sufficient evidence that the fish is safe to eat and does not differ nutritionally from other Atlantic salmon, the FDA has kept the company treading water for years, effectively condemning the commercial program to extinction. There are numerous other applications in various stages of R&D, including transgenic livestock with leaner muscle mass, enhanced resistance to disease and improved use of dietary phosphorous to lessen the environmental impacts of animal manure. (The fluorescent zebrafish was first developed as a means of detecting environmental pollution; it was engineered to fluoresce in the presence of certain Despite protests from activists, such as these on their way to demonstrate against the FDA’s Draft Risk Assessment on Clones at Washington, DC’s Capitol building last February, the FDA decided in January that such foods may be sold and consumed. FDA should now rethink its approach, stop its foot-dragging and promulgate a definitive, science-based policy on transgenic animals. N ew sc om .c om /B ill C la rk C O R R E S P O N D E N C E

US National Academies report misses the mark.

To the Editor: Last May, the US National Academy of Sciences (NAS) released its eleventh report since 1986 examining the safety and related policy issues of crops improved through biotechnology, commonly (if incorrectly) known as ‘GMOs’ (genetically modified organisms). This latest report1, Genetically Engineered Crops: Experiences and Prospects, comes at a particularly important juncture, when the Obama administration has stressed the importance of biotech innovation for the United States2 and acknowledged the need for regulatory policy to be coordinated and updated3. Unfortunately, the report not only contains several important inaccuracies and omissions, but also often fails to provide background necessary to understand the complex agricultural context and environment in which genetically engineered (GE) crops are adopted. Most important, it muddies the debate about yields of GE crops compared with ‘conventionally’ bred crops, gives undue credence and prominence to views backed by paltry peer-reviewed evidence, and provides precious little direction to policymakers on how to recalibrate the regulatory framework to emphasize science-based risk assessment and reduce discrimination against GE products compared with non-GE products. One glaring problem of the report is its failure to acknowledge widespread yield gains in multiple countries arising from better weed control with herbicide-tolerance technology, despite many reliable reports of this in the scientific literature4–11. The report examines yield and farm income effects in chapters 4 and 6, yet fails to refer to, or draw on, important and detailed peer-reviewed literature on the socioeconomic impacts of GE crops and many of the peer-reviewed literature cited therein. These omissions are difficult to understand, given that copies of some of these papers were sent to the NAS committee at an early stage of its review (G. Brookes and P. Barfoot, personal communication). In its discussion of yield gains associated with the use of GE technology, chapter 4 is particularly misleading. It cites several publications, including meta-analyses published in reputable, peer-reviewed journals, that draw on consistent findings from several studies12–14 and yet still finds a way to equivocate about these findings. The report presents unpersuasive alternative views and other possible reasons for yield differences based on a much more limited body of literature—literature that is often derived from research that is not reasonably representative of commercial farming practices. Some of these alternative explanations highlight valid examples where no or limited yield gains from using GE technology may be found, for example, in pest-resistant varieties in years where pest pressure is low. However, these situations do not reasonably represent what most farmers using GE crops have experienced over 20 years of use (otherwise, one would not expect the observed adoption rates and high repeat index). The issue of yield increases is complicated. There are myriad compounding variables, and it is difficult to sort and assign the proportion of observed yield increases that derive from genetic improvements (whether from conventional breeding or traits introduced with molecular genetic engineering techniques) versus, for example, improved agronomic practices and better equipment15. Some researchers have claimed an attenuation of yield increases in some crops before the advent of GE seeds, whereas others do not report such observations. But rather than grapple with this complexity, the NAS report obfuscates the issue in an odd way by focusing instead on the rate of change in the rate of yield increases with GE crops. The report seems to suggest that since genetic engineering has apparently not increased the rate of yield increases more rapidly than classic plant breeding, the actual demonstrable increases in yields delivered to farmers by GE seeds are inconsequential and can be ignored16. Farmers disagree, of course, and have cast their economic ballots for GE seeds at rates not seen with any other major innovation in the history of agriculture17. (Moreover, we would note that as they began to plant GE crops, many, perhaps most, farmers did not shift completely all at once, but compared conventional and GE side by side. In view of that, the high repeat index and farmers’ collective decision to expand their cultivation of GE crops is telling.) This unwillingness to overtly back GE crops, and the report’s efforts to give credence to alternative viewpoints—rather like the media’s obsession with giving two sides of an argument equal play, irrespective of which view is supported by the evidence—is puzzling. And it is also damaging for the following reasons. The politically correct insistence that “every opinion” counts extends to a recommendation that “governance authorities should actively seek public input on decisions, including decisions on how to approach emerging genetic-engineering technologies...and their regulation.” Although public dialog and engagement are among the many important facets for deciding which products are developed, it is less useful for the formulation of regulatory policy, particularly when complex issues of science and technology are involved. Indeed, experience has shown that public input on arcane scientific issues is not likely to be helpful. When the US The NAS report Genetically Engineered Crops: Experiences and Prospects was published earlier this year. C op yr ig ht T he N at io na l A ca de m ie s P re ss C O R R E S P O N D E N C E

Basic research is often best appreciated in retrospect.

volume 32 NumBeR 1 JANuARY 2014 nature biotechnology enzymology and nucleic acid chemistry that led to techniques for cutting and rejoining segments of DNA; advances in fractionation procedures that permitted the rapid detection, identification and separation of DNA and proteins; and the accumulated knowledge of microbial physiology and genetics, which pointed the way to a process by which ‘foreign’ DNA could be introduced into a cell’s DNA and made to function there. The result was the ability to move functional genes from one organism to another virtually at will—and it presaged the birth of modern biotech. Over the past 40 years, recombinant DNA technology has transformed numerous industrial sectors, including plant breeding and the production of pharmaceuticals and diagnostic tests. Its breakthrough drugs include human insulin, various treatments for cancer, cystic fibrosis, psoriasis, rheumatoid arthritis and certain monogenetic diseases, and vaccines that prevent certain infectious diseases. A further example is the creation of ‘hybridomas,’ hybrid cells made in the laboratory by fusing a normal antibodyproducing white blood cell with a cancer cell. This was originally done to combine desired features of each—namely, the immortality and rapid growth of the cancer cell and the ability of the normal cell to dictate the production of a single, specific ‘monoclonal’ antibody. The inventors had merely wanted to study the protein products of these fused cells to investigate the rates of cellular mutation and the generation of antibody diversity. It has turned out that these immortal, antibody-producing cells were useful not only for scientific inquiry but also as a novel kind of technologicaltherapeutic instrument for a variety of applications, including blockbuster drugs such as Rituxan (rituximab), Erbitux (cetuximab), Herceptin (trastuzumab) and Avastin (bevacizumab). Kirschner observes in his editorial that because of the “tendency to equate significance to any form of medical relevance,” research on nonmammalian To the Editor: The federal government expends vast amounts of money on various kinds of research, which run the gamut from investigations of fundamental physical and biological processes to applied research on what are judged to be national needs. Public funding for scientific investigations should generally be limited to basic research or proof-of-principle experiments—which would reasonably be defined as public goods—rather than efforts to extend science into marketable technologies or products. Research projects that receive federal funding should also follow recognized experimental methodologies and focus on nontrivial questions or problems. But funds are limited, and when one gets down to decisions on individual proposals, choosing among them may be difficult. The United States’ largest funder of biomedical research, the US National Institutes of Health (NIH; Bethesda, MD), several years ago introduced a new criterion for the ranking of grant proposals: “significance.” The NIH clarified the meaning of the term in an announcement: “Does the project address an important problem or a critical barrier to progress in the field? If the aims of the project are achieved, how will scientific knowledge, technical capability, and/or clinical practice be improved? How will successful completion of the aims change the concepts, methods, technologies, treatments, services, or preventative interventions that drive this field?” (http://grants.nih.gov/grants/peer/ guidelines_general/impact_significance. pdf). Reviewers of grants are required explicitly to express their opinions on the proposed research’s probable “impact,” defined as the likelihood that the proposed work will have a “sustained and powerful influence.” Correctly forecasting ‘significance’ and ‘impact’ can be a high bar, however, for basic research into the most arcane inner workings of cells, tissues and organisms in health and disease. As Harvard’s (Cambridge, MA) Marc Kirschner pointed out in a thoughtful and much-discussed June editorial in the journal Science (340, 1265, 2013), although we may be able to recognize sound science as it is being performed, “significant science can only be viewed in the rearview mirror.” Seemingly unrelated and obscure research areas may coalesce unexpectedly. After he received the 1969 Nobel Prize in Medicine or Physiology, my Massachusetts Institute of Technology (MIT; Cambridge, MA) microbiology professor, Salvador Luria, made a joke of the difficulty of perceiving the significance of one’s research findings at the time they are first obtained. To all of us who had congratulated him on the award, Luria sent a cartoon that showed an elderly couple at the breakfast table. The husband, reading the morning newspaper, says, “Great Scott! I’ve been awarded the Nobel Prize for something I seem to have said, or done, or thought, in 1934!”. A similar sentiment was conveyed eloquently in a 2011 editorial in Science (332, 767, 2011) by French biologist François Jacob in which he described the research that led to his own 1965 Nobel Prize in Physiology or Medicine. His lab was working on the mechanism that under certain circumstances causes the bacterium Escherichia coli to suddenly produce bacterial viruses, while at the same time another research group was analyzing, also in E. coli, how the synthesis of a certain enzyme is induced in the presence of a specific sugar. As Jacob wrote, “The two systems appeared mechanistically miles apart. But their juxtaposition would produce a critical breakthrough for our understanding of life”—namely, the concept of an ‘operon,’ a cluster of genes whose expression is regulated by an adjacent regulatory gene. Another quintessential example of both the synergy and the serendipity of basic research is the origin of recombinant DNA technology, the prototypic technique of modern genetic engineering. This resulted from the synergy among several esoteric, unrelated areas of basic research: Basic research is often best appreciated in retrospect correspondence

FDA is the wrong agency to regulate genetically engineered animals

VOLUME 35 NUMBER 7 JULY 2017 NATURE BIOTECHNOLOGY 23. Zetter, K. Wired https://www.wired.com/2015/06/hackers-can-send-fatal-doses-hospital-drug-pumps/ (8 June 2015). 24. US Food & Drug Administration. http://www. fda.gov/Safety/MedWatch/SafetyInformation/ SafetyAlertsforHumanMedicalProducts/ucm456832. htm (31 July 2015). 25. De Freytas-Tamura, K. The New York Times https://www. nytimes.com/2017/02/17/technology/cayla-talkingdoll-hackers.html?_r=0 (17 February 2017). 26. Smith, D. Radware Blog https://blog.radware.com/ security/2016/07/malware-and-botnet-attack-servicesfound-on-the-darknet/ (13 July 2016). 27. Mahmoud, R., Yousuf, T., Aloul, F. & Zualkernan, I. in 2015 10th International Conference for Internet Technology and Secured Transactions, 336–341 (2015). 28. Lohr, S. New York Times https://www.nytimes. com/2017/04/03/technology/trump-repeal-onlineprivacy-protections.html?_r=0 (3 April 2017). 29. Boudway, I. Bloomberg Business https://www.bloomberg.com/news/articles/2014-06-05/is-chris-dancythe-most-quantified-self-in-america (5 June 2014). 30. Wolf, G. https://www.wired.com/2009/06/lbnp-knowthyself/ Wired (22 June 2009). 31. De-Sola Gutiérrez, J., Rodríguez de Fonseca, F. & Rubio, G. Front. Psychiatry 7, 175 (2016). 32. Liu, C., Lin, S., Pan, Y. & Lin, Y. Medicine 95, e4068 (2016). 33. Montag, C., Duke, E. & Reuter, M. in Internet Addiction (Eds., Montag, C & Reuter, M). 131–139 (Springer International Publishing, 2015). 34. Demirci, K., Akgönül, M. & Akpinar, A. J. Behav. Addict. 4, 85–92 (2015). 35. Lanaj, K., Johnson, R.E. & Barnes, C.M. Organ. Behav. Hum. Decis. Process 124, 11–23 (2014). 36. Stothart, C., Mitchum, A. & Yehnert, C. J. Exp. Psychol. Human 41, 893–897 (2015). 37. Kushlev, K. & Dunn, E.W. Comput. Human Behav. 43, 220–228 (2015). 38. Montag, C. et al. BMC Res. Notes 8, 331 (2015). 39. Etkin, J. J. Consum. Res. 42, 967–984 (2016). 40. Javadi, A.H. et al. Nat. Commun. 8, 14652 (2017). 41. Czikszentmihalyi, M. Flow: The Psychology of Optimal Experience (Harper & Row, New York, 1990). 42. Centers for Medicare and Medicaid Services. http:// www.cms.gov/Regulations-and-Guidance/Legislation/ EHRIncentivePrograms/Downloads/EHR_Medicare_ Stg1_BegGuide.pdf (April 2014). behavior, technologies could include periods of time where its quantification functionality is unavailable. Technology could be programmed to initiate such ‘blackout’ modes on a daily basis, freeing up time and attention to focus on experiences in a more immediate manner. Devices would continue to track measurements during this period; the blackout mode would simply prevent the user from easily accessing her quantified data, thereby allow for daily periods of unquantified existence.