The illusion of control in germline-engineering policy

VOLUME 35 NUMBER 6 JUNE 2017 NATURE BIOTECHNOLOGY correlation between gCSI and CTRP data sets is significantly increased using GRAOC rather than AUC as a response metric (P = 1.3 × 10-3, Student’s t-test; Supplementary Fig. 4). And third, efficacy as measured by GRmax and potency as measured by GR50 differ at a biological level, carry complementary information (low mutual information), and are associated with largely non-overlapping genetic alterations. In principle, variation in potency and efficacy can be captured by integrating across dose–response curves (GRAOC), but we find that information content is maximized if GR50 and GRmax are considered independently. Because the ultimate purpose of antineoplastic drugs is to kill cancer cells13, and high potency is no guarantee of good efficacy, we propose that the best drugs and most important pharmacogenomic associations are not those associated with low IC50 values, but rather those that result in the most negative GR value at clinically relevant drug concentrations (e.g., Cmax). Relating in vitro measures of drug sensitivity to in vivo responses remains challenging17, but for this to have any chance of success it is essential that in vitro data are as informative and reproducible as possible. Figure 2 Variation in efficacy and genomic enrichment based on GRmax values in the gCSI data set. (a) Distribution of GRmax and GR50 values for each drug across all cell lines. GR50 values are capped at 31 mM. (b) GR50 and GRmax enrichment by genomic alteration for each drug and tissue. Numbers represent significant (FDR < 0.15) associations for GRmax alone (red dots), GR50 alone (blue), or GRmax plus GR50 (purple). (c) Distribution of GRmax values for docetaxel in ovarian cancer lines based on BCL2 deletion status. Rank-sum P-value is reported. (n = 23). -1 -0.5 0 0.5 1

correlation between gCSI and CTRP data sets is significantly increased using GR AOC rather than AUC as a response metric (P = 1.3 × 10 -3 , Student's t-test; Supplementary Fig. 4).
And third, efficacy as measured by GR max and potency as measured by GR 50 differ at a biological level, carry complementary information (low mutual information), and are associated with largely non-overlapping genetic alterations. In principle, variation in potency and efficacy can be captured by integrating across dose-response curves (GR AOC ), but we find that information content is maximized if GR 50 and GR max are considered independently.
Because the ultimate purpose of antineoplastic drugs is to kill cancer cells 13 , and high potency is no guarantee of good efficacy, we propose that the best drugs and most important pharmacogenomic associations are not those associated with low IC 50 values, but rather those that result in the most negative GR value at clinically relevant drug concentrations (e.g., C max ). Relating in vitro measures of drug sensitivity to in vivo responses remains challenging 17 , but for this to have any chance of success it is essential that in vitro data are as informative and reproducible as possible.  The illusion of control in germlineengineering policy To the Editor: The arrival and rapid adoption of the clustered, regularly interspaced, short palindromic repeats (CRISPR)-CRISPRassociated protein 9 (Cas9) system 1 has sparked ethical and societal controversy around genome editing of the human germline. Here, I point out the fallacy that such technologies and their applications can be globally prohibited on the basis of universal ethics and bans-the so-called 'illusion of control' . A look at previous technological developments suggests instead that differentiated and multi-faceted approaches that take into account the broadest range of possible ethical and social issues would be preferable for the oversight of CRISPR-Cas9 germline engineering. Such an approach would not only be more effective but also ensure that society has the greatest chance of capitalizing on potential opportunities of the technology. To date, the CRISPR-Cas9 technique has been used to edit the genomes of (nonviable) one-celled human embryos 2,3 . The first of these genome-editing studies international scientific consensus against reproductive cloning due to its low efficiency and the unacceptable potential health issues associated with cloned offspring-no binding global convention has been agreed upon. What's more, the resulting non-binding United Nations (UN; New York) "Declaration on Human Cloning" remains ambiguous by calling to "prohibit all forms of human cloning inasmuch as they are incompatible with human dignity and the protection of human life" 13 .
The ambiguity in the UN declaration arises from its failure to define or interpret the forms of cloning that are incompatible with human dignity or what human life and human dignity should mean. The reason this ambiguity exists was because UN negotiations had to address the issue that dignity and life have different meanings in different cultures and religions globally 14 . In line with this, regulations vary in different countries or regions of the world, ranging from prohibition of all forms of cloning, to selectively permitting therapeutic cloning, to no official regulations at all (Table 1). Furthermore, the Council of Europe's (Strasbourg, France) legally binding "Convention on Human Rights and Biomedicine, " which prohibits in its protocol on cloning the creation of "a human being genetically identical to another human being, whether living or dead, " has not been signed or ratified by various member states, although it explicitly leaves the interpretation of "human being" to national policies to allow therapeutic cloning where it is accepted 15 . That (reproductive) cloning appears not to have been performed anywhere so far, including in countries where there is no legal ban, may encourage supporters of such international declarations or conventions. However, it can hardly provide evidence that they were instrumental in-or even contributed to-this outcome, especially given the huge technical hurdles that reproductive cloning faces 16 .

Illusion of (universal) control
The idea that biological technologies, like CRISPR-Cas9, or their applications can be (continuously) prohibited, and that this can be done globally thus appears to be reminiscent of the 'illusion of control' in psychology-a phenomenon in which people's beliefs in the control are greater than can be actually justified. For instance, people act as if they have control in situations that are actually determined by chance 17 . Similarly, when people envision that they will obtain a certain result that they then achieve, published in the spring of 2015 prompted debate from within the field and by national academies about not only safety issues (such as off-target effects) but also ethical-moral questions. As a result, calls and statements have been released urging researchers to cease performing CRISPR-Cas9 experiments on human germ cells in a clinical context until safety issues have been resolved and the ethical and social implications of directed germline modifications broadly discussed among diverse societal groups and the public 4-6 . In February 2017 a report by the US National Academies of Sciences, Engineering, and Medicine concluded that clinical trials might be permitted and outlined criteria for a strict regulatory framework. These include the restriction to preventing a serious disease, credible pre-clinical and/or clinical data on risks and potential health benefits, or long-term, multigenerational follow-up (that still respects personal autonomy) 7 . Furthermore, several groups have suggested the need to develop norms and harmonize regulations internationally or globally 5,6 .
Among the numerous ethical and social challenges of genome editing and its potential applications, a key unresolved issue relates to the notion that certain technologies or applications can be prohibited globally. The notion of banning technologies or applications worldwide has come to the fore as calls have grown for regulating germline editing in humans. The central assumption underlying this notion is that technologies can be prohibitedcontinuously or until they are 'safe enough'-and, moreover, that this can be done globally.
Considering reproductive technology as a whole, history provides several examples of procedures that prompted controversy due to ethical (including safety) or moral concerns once they became technically feasible. These include the use of assistedreproduction techniques (ARTs), such as in vitro fertilization (IVF), intracytoplasmic sperm injection (ICSI), and preimplantation genetic diagnosis (PGD). As time has passed, these technologies have become more and more widely adopted and accepted across the globe. Today, they have come into use, even in countries where they were initially prohibited for considerable time periods (e.g., Costa Rica, which only effectively lifted its ban on IVF in 2016; and Germany, which only granted conditional permission for PGD in 2011).
In 2015, the United Kingdom approved therapeutic approaches that involve three-parent mitochondrial genome replacement (MGR) in egg cells or one-cell embryos, which thus entail inheritable changes to the human germline. Furthermore, in the Ukraine, MGR has been applied to treat infertility, with a first baby girl reported to have been born early this year 8 . Similarly, US fertility researchers recently announced that they had conducted MGR in Mexico to prevent mitochondrial disease in a boy born last April 9 .
An expert panel recently convened by the US Food and Drug Administration (FDA) concluded that clinical investigations of MGR are ethically permissible, though only under strict conditions, for example, if restricted solely to male embryos, which cannot pass on mitochondrial DNA to later generations 10 . Yet, by the time this article went to press, MGR therapy was still effectively banned in the United States owing to an FDA decision in 2001 (http:// bit.ly/2q2898X) because of concerns over safety issues such as negative effects from mitochondrial heteroplasmy and mitochondrial-nuclear mismatches. The decision was made in response to ART applied in several fertility clinics that involved the transfer of egg cytoplasm (including mitochondria) 11 . The decision indicates that the use in therapy of human cells "involving the transfer of genetic material by means other than the union of gamete nuclei" requires application to the FDA for permission.
Even the idea that technologies can be banned or prohibited until they are proven by research to be 'safe enough'-at least in the context of reproduction and/ or therapies-appears to be disputable. What safe enough means may in practice be as much a question of perception involving different values and interests (and weighing perceived benefits against perceived risks) as it is one of available 'strictly scientific' evidence or numbers (should they be available at all before first clinical applications). An example may be the decisions for the first ART via ICSI by four couples after 3-13 years of unsuccessful trials with traditional ART methods, even though only data from rabbits were available 12 .
An ambiguous state of global union Imposing legal regulations that might prohibit the use of ART technology globally also appears non-feasible, at least when ethical or moral arguments stand in the way of new medical opportunities. Even with respect to human somatic cell nuclear transfer-and despite a seemingly broad C O R R E S P O N D E N C E modification" 23 (see also ref. 24). The bill may thus also affect the FDA's capability to review MGR therapy approaches (see above).
When it comes to developing regulations on whether or under which conditions germline modifications should be permitted, entrenched moral stances and views on ethics will always be confronted by, and weighed against, the hopes and needs of affected people and their families and caregivers. These may include a desire for people to have their 'own' (genetically related) children not suffering the same disease (e.g., deafness or cystic fibrosis) as both parents do, or new fertility treatments that may be (felt to be) needed due to economic or cultural factors. Naturally, the result of the confrontation between societal ethics and individual hopes-be it a strict ban on all forms of human germline interventions or more differentiated they frequently overestimate their influence in bringing about the result 18 .
It is very unlikely that a global ban across jurisdictions is feasible. Indeed, current regulations related to germline interventions vary considerably across the globe 19,20 . They encompass bans-based on law or lessenforceable governmental or research council guidelines-that prohibit both research and clinical or reproductive applications; or that prohibit clinical or reproductive application only; or that remain sufficiently unclear to raise doubts as to whether they cover human germline gene modification at all ( Table 1).
In the United States, no federal law describes an outright ban on human germline interventions and work on human embryos, and in some states researchers could do research on embryos with private funding (as was the case with MGR experiments in Oregon in 2012; ref. 21). Yet, there are regulatory means in place that in fact prevent human germline modifications in publicly funded research and in clinical settings. Thus, the US National Institutes of Health does not fund "any use of gene-editing technologies in human embryos" and its Recombinant DNA Advisory Committee (RAC) will not entertain or review proposals for human germline alterations 22 . Furthermore, the FDA has regulatory authority over cell and gene therapy applying to any (not only publicly funded) research and clinical trials in the United States, and the agency has never approved a proposal to modify the germline.
In addition, a renewable provision (that has been extended twice into 2017) of the US Congress's Fiscal Year 2016 omnibus spending bill prevents the FDA from using the budget to evaluate or permit trials with human embryos "intentionally created or modified to include a heritable genetic The Convention's protocol on cloning prohibits anyone "to create a human being genetically identical to another human being, whether living or dead." But it explicitly "leaves the domestic laws of the States to define the scope of the term 'human being'" (http://www.coe.int/en/web/conventions/full-list/-/conventions/ treaty/168). Furthermore, it demands that "an intervention seeking to modify the human genome may only be undertaken for preventive, diagnostic or therapeutic purposes and only if its aim is not to introduce any modification in the genome of any descendants" (Article 13); and that "where the law allows research on embryos in vitro, it shall ensure adequate protection of the embryo" (Article 18). b However, the FDA has regulatory authority over cloning technology and over cell and gene therapy. Furthermore, research using stem cells derived from human cloning and gene-editing technologies in human embryos are not eligible for government funding. c Various states have passed laws that prohibit or permit different forms of cloning or embryo research, whereas some states have no policy or legislation on embryo research and human cloning 30 (http://www.ruf.rice.edu/~neal/stemcell/Sup_Info.html). d Though there is no federal law that bans outright genetic modification of the germline, a provision of the Consolidated Appropriation Act (omnibus spending bill) of 2016 prevents FDA from using federal funds to evaluate and/or permit trials related to human embryos "intentionally created or modified to include a heritable genetic modification" 23 .
Ambiguous: indicates that it remains unclear from policy documents whether a practice is banned or may be permitted (e.g., if definitions of key terms are left open); banned: signifies that a practice is prohibited (explicitly or implicitly) by national law or guidelines; permitted: indicates that a practice is either allowed (but not regulated) by laws or guidelines or that it is allowed and regulated (e.g., requiring case-by-case licensing by authorities); no policy: denotes that a practice is not considered in policy documents; ?: unknown or unclear whether this practice is addressed by national policies. questions, and thus improve policies over time-based on experiences (about hardly predictable interactions between society and technology) from different conditions and involving a wide variety of perspectives 29 . Such deliberation processes may reduce the danger of missing potentially important ethical, social or political implications, and further strengthen mutual trust. 'Illusion-of-control'-like notions of universal ethics and bans on germline applications may thus favor one-dimensional thinking and illusionary policies, harboring the danger of leaving aside potentially more realistic and efficient policy alternatives. The latter type of 'polycentric' approach should be better capable of exploring the opportunities of genome-editing techniques, foster collaboration between countries and enhance the chances to grasp the broad range of possible ethical and social issues that would have to be taken into account by efficient and trust-building policies. They may also require governments, non-profits, investors and corporate managers to invest more effort to understand and experiment with conditions that both foster and ethically guide innovation.
Ensuring that these differentiated and polycentric policymaking strategies take into account the realities of 'manyness' (i.e., societal differences in values and interests across the globe) would mean tailoring regulations around specific conditions rather than rejecting applications of technologies as a whole. Such an approach to regulations would not only allow space for experimental efforts that generate important scientific data (e.g., on safety issues), but also reduce incentives for researchers to move to countries with more lax standards or less stringent oversight. At the same time as they allow research to progress, they would also allow meaningful democratic deliberation processes on futures with genome-editing applications that people may, or may not, want.

Harald König
Karlsruhe Institute of Technology, Institute for Technology Assessment and Systems Analysis (ITAS), Karlsruhe, Germany. e-mail: h.koenig@kit.edu regulations-will strongly depend on the prevalent ethical and moral stances of the societies in which those individuals are found. And these societal ethical or moral stances will differ across the globe. As in the case of cloning, moral stances with respect to the compatibility with human dignity may also be a key issue with respect to germline modification.
The UN Declaration on Human Cloning includes statements on human genetic engineering. But as for its language about reproductive cloning, it remains vague and ambiguous with respect to genetic modification and human dignity, calling for the adoption of "measures necessary to prohibit the application of genetic engineering techniques that may be contrary to human dignity" 13 . Notions of global bans, and maybe even of globally harmonized regulations, on germline interventions-at least of ones that would not be too vague or ambiguous-may thus in fact include features related to the 'illusion-of-control' .

Alternatives to illusions
Features like exaggerated belief in control or in the causal role of one's own intentions, or unrealistic optimism are elements of normal human thought. They can promote motivation, persistence or the ability to care for others 25 . However, when it comes to the development or choice of policies, it may be worth considering that these phenomena could also have a downside: they may entail less 'realistic' and thus less-efficient policies. Thus, policies may be followed that strive to universally, and maybe continuously, ban or prohibit technologies or their applicationirrespective of their purpose-on a national level or even globally. But on the basis of the evidence, there is in fact little or no empirical support that such strategies work.
In addition, they may obscure and distract from alternative policy approaches, focusing on processes that we could better or more realistically control. Such processes could be case-by-case evaluations, dealing with specific applications in which we use knowledge and the technologies in question, such as CRISPR-Cas9 genome editing. Developing and delineating tests and standards (including clinical ones) for techniques or therapeutic tools used in the specific context of these applications may be important elements of such evaluations. Furthermore, focusing on concrete applications and conditions should facilitate interpretations and definitions (including legal ones) related to applications and ethical boundaries 20 . In contrast to global regulations or bans based on efforts to 'globalize' ethics that universally disapprove germline applications of genome editing, alternative, more realistic policies would support more case-oriented approaches.
Such policies may in fact benefit from experiences coming from diverse governance schemes or authorities. These may involve state, regional and national regulations, multiple monitoring bodies (both public and private), but also formal and informal bilateral or multilateral interactions and the engagement of stakeholders on various levels. Stakeholders should include not only researchers and ethicists, but also companies and investors, non-governmental organizations, groups of concerned citizens or families who would be involved in deciding, for instance, on (experimental) approaches.
The difficulty of deciding who should be included in such consultations should not be underestimated. For example, it may be important to engage in the policy development processes of not only do-ityourself biology and biohacker communities as well as biohacker spaces, but also funding sources, such as crowd-funding platforms or investor-supported 'hackubators' , even though human germline applications of CRISPR-Cas may not (yet) appear as a goal for these groups 26 . Involving them and their efforts on codes of ethics may be beneficial, not least of all because any bans or guidelines on institutional research or research funding may hardly provide guidance and oversight on such noninstitutional activities 27 .
These more 'polycentric' patterns of approaches and stakeholders may promote key elements linked to polycentric governance in other contexts, like diverse experimental efforts, enhanced collaboration or mutual monitoring and trust 28 . These elements could be important to facilitate bilateral and multilateral exchange between nations, and gain scientific data and knowledge on safety issues by carefully crafted and monitored laboratory and clinical experimental approaches. But they may also prove beneficial to the development of democratically meaningful deliberation processes. These would need to go beyond the type of expert-dominated summits or one-time public consultations with largely expert-driven agendas (that resulted from calls, including for global action, and statements by various groups 4-6 so far). Furthermore, these processes would need to contribute to understanding risks by continuously revisiting earlier choices and