Drugs, genes and screens: The ethics of preventing and treating spinal muscular atrophy

Abstract Spinal muscular atrophy (SMA) is the most common genetic disease that causes infant mortality. Its treatment and prevention represent the paradigmatic example of the ethical dilemmas of 21st‐century medicine. New therapies (nusinersen and AVXS‐101) hold the promise of being able to treat, but not cure, the condition. Alternatively, genomic analysis could identify carriers, and carriers could be offered in vitro fertilization and preimplantation genetic diagnosis. In the future, gene editing could prevent the condition at the embryonic stage. How should these different options be evaluated and compared within a health system? In this paper, we discuss the ethical considerations that bear on the question of how to prioritize the different treatments and preventive options for SMA, at a policy level. We argue that despite the tremendous value of what we call ‘ex‐post’ approaches to treating SMA (such as using pharmacological agents or gene therapy), there is a moral imperative to pursue ‘ex‐ante’ interventions (such as carrier screening in combination with prenatal testing and preimplantation genetic diagnosis, or gene editing) to reduce the incidence of SMA. There are moral reasons relating to autonomy, beneficence and justice to prioritize ex‐ante methods over ex‐post methods.

tions to DNA sequences. Laboratory experiments have shown that CRISPR-Cas9 can correct pathogenetic mutations at the embryonic stage. 8 While germline gene editing has been used reproductively in one case, 9 more research is needed before such techniques can be considered safe enough for a clinical setting. However, in the future, it is plausible that they could be used to correct the mutations that lead to SMA. Such technologies could be used to edit embryos produced through IVF so that they contain a functional copy of the SMN1 gene, or they could be used pre-conceptionally to alter sperm or egg cells.
This multitude of treatments and prevention options for SMA raises questions regarding treatment prioritization. When there are multiple therapeutic approaches to SMA available to public health systems, which ought to be funded and supported? 10 In this paper, we investigate the ethical aspects of treatment prioritization for SMA. To help structure our discussion, we distinguish between 'ex-post' and 'ex-ante' approaches to the treatment and prevention of SMA. Ex-post interventions are those that are administered after the diagnosis of an infant with SMA (such as nusinersen and AVXS-101), whereas ex-ante interventions are those that prevent cases of SMA from occurring in live-born infants. We argue that individuals and states have moral reasons to pursue both ex-ante and ex-post approaches to treating SMA. In some contexts, we should prioritize carrier screening (which enables ex-post approaches) over treatments like nusinersen. We show that there are moral reasons of autonomy (Section 3), beneficence (Section 4) and justice (Section 5) to prioritize carrier screening. We begin by discussing the disability rights criticism of genetic screening and applying it to the specific case of SMA.

| D ISAB ILIT Y RI G HTS CRITI QUE S AND S MA
A prominent strand of criticism of reproductive technologies that prevent disability comes from the disability rights community. 11 Selection against disability is morally problematic, according to these views, for at least two reasons: it expresses morally problematic and hurtful attitudes, and it exemplifies an intolerance to diversity.
The first reason is often discussed under the banner of 'The Expressivist Argument'. 12 The Expressivist Argument claims that when people choose not to implant an embryo, based solely on the fact that it is predisposed to develop a disability, they express an immoral or otherwise objectionable attitude towards existing disabled individuals. Specifically, they express the judgment that the disabled person is worse or in some way less worthy of existence than other persons. Such judgements are representative of challenges disabled persons face in their daily lives. Through the act of selection, parents make a judgement about an entire embryo (that it is not as preferable as other embryos), solely on the basis of the knowledge that it is predisposed to develop a disability. This mirrors the type of discrimination disabled people are exposed to, where the mere knowledge that a person has a disability is taken to imply facts about the individual as a whole.
In contrast, treating a person who is disabled does not express these same problematic attitudes. Treatment does not involve a judgement that an individual as a whole is more or less worthy of existence than others -it may simply represent the belief that an individual would benefit in some way from a particular intervention.
The second line of criticisms from the disability rights community to screening is that it demonstrates an intolerance to diversity. This argument can be pursued in at least two ways. First, it can be argued that the selection against disability shows a negative attitude towards embracing diversity in the family. When parents select against an embryo, based solely on the knowledge that it is predisposed to develop a disability, they demonstrate a problematic attitude towards parenthood, namely that they will only love their child if it has certain traits and characteristics. Good parents should embrace the diverse possibilities that children give rise to. In the words of Michael Sandal, they should be 'open to the unbidden' 13 . Selection against disability is incompatible with these attitudes. Furthermore, selection against disability might be symbolic of a problematic attitude to diversity in society. Selection against disability can be taken to imply that society would be better off in the absence of disability. But such views are challenged by at least two considerations. The first is that being disabled can be a source of different perspectives that strengthen society rather than weaken it. 14 Disability is a source of narrative resources, as the lives and struggles of disabled people provide a counterpoint to abled-body lives and increase our understanding of what it means to be human.
Disabilities also provide epistemic resources, because the experience of being disabled provides distinctive 'ways of knowing' and makes possible new forms of aesthetic expression and evaluation. 15 Relatedly, disability also provides a cognitive resource, as the experience of being disabled leads one to develop distinctive perspectives and heuristics. 16 Recent empirical work has shown that cognitive diversity can be more important than individual ability for our collective capacity to solve complex problems. 17 The presence of disabled people in our population can thus have wide social benefits. 18 Secondly, it is argued that many of the negative aspects of disabilities can be seen as socially constructed rather than as being objective features of the condition itself. 19 If society were arranged differently, the argument goes, those with disabilities would experience fewer disadvantages associated with their condition, and parents would not be motivated to select against disabilities.
In sum, the disability critique of genetic screening gives us reason to be more sceptical of using PGD than ex-post approaches.
Interestingly, these views can also be interpreted as implying that the use of gene editing is less problematic than the use of genetic selection to avoid disability. Whereas genetic selection expresses a view that some embryos are more worthy of existence than others, gene editing potentially involves modifying a single embryo. It is much more analogous to standard treatment than genetic selection. 20 The conceptual basis of the disability critique, as well as its force and scope, have been challenged. 21 Most notably, many argue that selection against disability need not express a negative opinion about disability or disabled people at all. 22 Rather than engaging with these debates here, we will simply note that disability critiques apply less clearly to conditions like SMA1, which cause death in early childhood.
There are no adults with SMA1 who could be directly harmed by the attitudes expressed by testing and selection against SMA1.
It might be countered that selection against SMA1 expresses negative attitudes towards those with milder forms of SMA. But here it seems that we can point to the very obvious difference between  For the same reasons, selection against SMA1 does not symbolise a problematic attitude to diversity. Differences that result in childhood death are certainly not the type of difference we believe that parents should be indifferent to and embrace as part of their 'openness to the unbidden'. Far from being a source of novel, narrative, epistemic and cognitive resources, SMA1 deprives those affected from developing a unique way of being.
For these reasons, we must look beyond the disability rights critiques of selection to help us decide how to prioritize different interventions that can reduce the disease burden of SMA.

| AUTONOMY
Autonomy is a rich, multifaceted concept that is ubiquitous in philosophy and medicine. The term strictly translates to self-rule and was originally used by Ancient Greek philosophers to refer to a property of cities. 23 Greek city-states were referred to as 'autonomous' if they were governed by laws that were of their own making, rather than having laws imposed on them by external powers. Since the 18th century, autonomy has increasingly been applied to individuals rather than collectives. Drawing on the traditional understanding of autonomy, Kant argued that for a person to be autonomous, they had to live in accordance with moral principles that were of their own making, rationally derived and personally endorsed. 24 Today, the notion of autonomy is increasingly used in a less formal sense to signify something like the ability of persons to make decisions without outside interference. 25 Rather than take a stand on how best to understand autonomy, we will instead focus on two values that are promoted by individual autonomy and which can be seen as independently important in medicine: sovereignty and non-alienation. 26 A concern for sovereignty reflects the fact that some decisions are simply the domain of individuals, regardless of how they will exercise this capacity. The idea of sovereignty can be clearly understood through examples that involve property rights. If you own a car, you have sovereignty over it. It is your decision whether a friend can borrow it, no matter how strong your friend's moral claims might be to borrow the car (e.g. they want to drive someone to the hospital). Decisions about the car are simply yours to make, as you have sovereignty over it. Similarly, the concept of 'bodily sovereignty' reflects the idea that people have innate sovereignty over their body and have the ultimate say over decisions that concern it. In reproductive medicine, sovereignty is most commonly associated with the Having a child with a serious disability would significantly affect how one's life would go. Although raising a child with a disability can be highly rewarding, and give rise to different types of parenting experiences, it is often a very different experience to raising a non-disabled child. A concern for non-alienation suggests that people who are having a child should be free to influence whether their child has a disability or not, and make this choice Such considerations also suggest that parents have an interest in choosing to have a disabled child. We take no stand on this issue in this paper. We note though that other moral considerations discussed below (beneficence and justice) might count against this. available to parents in different circumstances. This implies firstly that we have reason to make carrier testing available to couples thinking of having a child. The first major decision facing most parents who are at risk of having children with SMA is whether to conceive through IVF and use PGD, to conceive naturally and undertake prenatal testing and pregnancy termination, or to conceive naturally

| B ENEFICEN CE
The principle of beneficence is used widely in medicine and philosophy. 32 As a principle of medical ethics it is often used to denote 'the doing of good, the active promotion of good, kindness and charity', 33 or the moral obligation 'to act for the benefit of others, helping them to further their important and legitimate interests, often by preventing or removing possible harms'. 34 How we should understand beneficence in a medical context is a complex issue. It partly depends on how one understands the goals of medicine. According to some views, the goals of medicine should be to promote health and to cure disease. 35 Others argue that med-icine should have a broader role, in supporting the overall well-being of patients. 36 Complicating matters is that neither health nor well-being has an accepted definition-and indeed health is sometimes defined as a state of well-being. 37 It is beyond the scope of this article to delve into any of these issues in any detail. Rather, we will adopt a general view that beneficence is supported by increasing the years of life and quality of life.
Both these factors are regularly used as a guide to medical practice. 38 While 'quality of life' is itself a contested term and can be understood in a multitude of ways, 39 these specifics will matter little when considering whether beneficence provides moral reasons to distinguish between different interventions for SMA. Who benefits from this intervention? Of course, Beth and Jerry will benefit, as their preference to have a child free from SMA is fulfilled-but does Craig benefit? According to one understanding of benefit, 'an action benefits a person only if it makes her better off in some respect than she would have been, had the action not been performed'. 40 Imagine if Beth and Jerry do not conceive through IVF. Does this leave Craig in a worse off state?
No, rather he will not exist at all. Craig himself does not benefit from the action of his parents for there is no scenario where Craig exists and has SMA. In contrast, when we treat a child with nusinersen, or some other ex-post therapy, we make that child better off than they would be if treatment were withheld. Initially, then, considerations of beneficence seem to favour ex-post over exante interventions.
However, there are different ways to understand the requirements of beneficence. We could concede that PGD does not benefit 'Craig', but rather benefits Beth and Jerry's 'next child', whoever that particular person may be. 41 The relevant comparison to make is between Craig's well-being and the well-being of the child Beth and Jerry would have had, had they not engaged in PGD. In cases where that next child would have had SMA, PGD confers a significant benefit. This way of understanding beneficence denies that there is 'a counterfactually better-off condition on benefiting', 42 so we only benefit others if we make them better off in some sense than they would have otherwise been. As has recently been argued by Molly What about prenatal testing and termination? As no child is directly produced through termination, at one level no one benefits.
Prenatal testing and termination may, therefore, be the least beneficial treatment. However, this neglects the fact that many parents who engage in prenatal testing and terminations will likely try to have another child without SMA and have this explicit motivation when they choose this option. Each subsequent child the couple has will have a 25% risk of having SMA1. However, in cases where termination leads to another child who is SMA-free, it is, arguably, the equivalent to PGD being used to prevent SMA.
This suggests that in cases where couples are likely to be able to conceive more children, prenatal testing and termination is more beneficent than ex-post treatments like nusinersen. However, in cases where the couple is unlikely to be able to have another child, it becomes the least beneficial option.
In the future, gene editing may provide a way of completely curing SMA at the embryonic stage. Again, imagine Beth and Jerry, both of whom carry an SMN1 mutation. They wish to avoid having a child with SMA and conceive through IVF. This time, they produce a single embryo, Karen, who has two SMN1 mutations. Gene editing is used to correct these mutations. Karen has a normal life expectancy and no disability.
Gene editing is thus like PGD, in that it results in a baby without SMA1 and, all else being equal, a normal life expectancy. Hence it is a highly beneficial option. 46 Of course, while gene editing has risks of off-target mutations, genetic selection is in one way safer and preferable. 47

| J US TI CE
The Roman lawyer Cicero gives one of the earliest definitions of justice as 'the virtue which assigns to each his due'. 48 This broad definition still captures the core concerns of justice today. While justice encompasses many elements of ethics and law-such as the punishment of crimes, the distribution of resources, and the relationship between society and individuals, it fundamentally represents a concern for giving people what they are 'due'. In moral philosophy, justice is seen as a particularly powerful value. If something is labelled as 'unjust'-be it an institution, policy, or individual action-this generates strong, if not decisive, reason to reject it. 49 | 499

GYNGELL Et aL.
We can distinguish between different domains of justice, including distributive justice (how to distribute resources justly), corrective justice (how to compensate people for past wrongdoing done to them) and procedural justice (how people are treated to establish rules and regulations).
In medicine, questions of justice usually focus on distributive justice. Medical resources, such as drugs, specialist time, pathology tests etc., are limited. The demand for these resources outstrips their availability. This raises the question of how to allocate them in a way that 'assigns each their due'.
How should we distribute limited medical resources, such as access to SMA treatments? Such questions have mattered very little until the last few decades. Historically there were few treatments for most rare diseases, and certainly not multiple effective treatments for a single catastrophic condition. However, advances in medicine and technology mean that states must now make decisions about which treatments to develop and fund, and therefore which goals they should pursue when distributing medical resources.
As we saw in Section 3, a simple metric that can be used to compare alternative treatments is their effect on survival. One view, then, is that a just distribution of medical resources would be the distribution that improves the life expectancy of a population the most.
However, as we saw in Section 3, there are problems with metrics that focus merely on survival. People care about more than just surviving: they care about what they can do with that survival. Any theory of the just distribution of medical resources must take this into account. QALYs). While, such empirical work is highly controversial, these difficulties do not affect our subsequent analysis. 52 Increases in QALYs essentially track increases in quality and years, which we discussed in Section 3. As we saw, ex-ante interventions that prevent SMA can be expected to generate more well-being than ex-post interventions that treat it after the diagnosis. Accordingly, they also generate more QALYs.
However, when considering justice and not beneficence, we must look beyond the amount of QALYs that an intervention generates, to its cost-effectiveness. Cost determines over a population with limited resources how many people can be treated.
When we approach issues through the lens of justice, considerations of the cost of treatments are not merely practical considerations, but also key ethical issues. In public health systems, there is a limited amount of resources. If one individual is treated with an expensive treatment when they could be treated with a cheaper treatment, that reduces the resources available to others. This indirectly harms others who could use this resource.
There is an ethical imperative to use public health resources in a cost-effective way. The cost of treatment is, therefore, an ethical consideration.
As discussed in the Introduction, nusinersen is expected to be very expensive: US$750,000 for the initial treatment, and US$375,000 for each year of subsequent treatments. Let us suppose that the public health system fully funds it, and it manages to extend a patient's life for 10 years. This means that the course of treatment will cost the health system US$4.125 million. In contrast, PGD will cost approximately US$15,000. If a couple uses PGD rather than nusinersen, this will save over US$4 million of health resources that could be used to help others. Furthermore, PGD generates many more QALYs, as we argued in Section 3. Favouring PGD over nusinersen thereby ensures that more people get access to medical resources, which in turn promotes justice.
The cost of prenatal testing and termination is even cheaper than PGD, with a total cost of approximately US$5000. When this is likely to result in another child, it also generates more QALYs and results in an even more just distribution of resources (assuming the child is SMA-free).
While the cost of gene editing is unclear at the stage, it will likely be in the range of PGD and IVF, perhaps around US$30,000.

| Screening
The above analysis of the cost-effectiveness of PGD and prenatal testing overlooks a central issue. In our discussions, we have been assuming that couples know that they are carriers for SMA, and thus prenatal testing and PGD are visible options to them. However, this will not be the case for the vast majority of SMA carriers. Only people who have had a DNA carrier test will know with certainty whether or not they carry an SMN1 mutation, and this will mostly only be if they have a family history and pay for and access carrier screening tests. found that the cost was US$4.9 million per QALY gained. 53 However, this study has been criticized for looking only at QALY gain by the mother of an affected child, 54 and thus overlooking the fact that screening also generates QALYs through the existence of SMA-free children who would not exist without the screening programs.
Furthermore, the last few years have seen a vast improvement in the cost-effectiveness and breadth of so-called 'next generation' screening methods, which can detect carrier status for up to 400 conditions in one test. A study looking at the cost-effectiveness of population carrier screening for 176 conditions (including SMA) estimated that such programs would cost ~$50,000 per QALY gained. 55 The relative advantages of population carrier screening combined with PGD are particularly clear when we consider the fact that nusinersen and AXVS-101 are most effective when administered in pre-symptomatic infants. 56 To maximize the efficiency of these treatments, they will need to be combined with newborn screening, to allow infants with SMA to be identified and treated quickly. 57 In July 2018, The U.S. Department of Health and Human Services Secretary recommended the inclusion of SMA in the national newborn screening panel, in part to facilitate affected infants being able to access these new therapies. 58 Newborn screening would add a substantial additional cost, making these treatments even less cost-effective.
In sum, we have reasons of justice to ensure that medical resources are equitably distributed. This requires that we consider the cost-effectiveness of potential medical treatments. When people know that they are SMA carriers, there are strong reasons for justice to promote ex-ante intervention like PGD and prenatal testing and termination, over ex-post methods like nusinersen. While more empirical work needs to be done to assess the cost-effectiveness of carrier screening, especially for SMA when this is combined with screening for other Mendelian disorders, it appears highly likely that such programs should also be promoted on justice grounds. There are also strong distributive justice arguments to make carrier testing, PGD and IVF, or prenatal testing and selective termination free of charge to affected couples.

| CON CLUS IONS
A plethora of new interventions promise to reduce the burden of SMA for families. However, they raise difficult questions regarding resource prioritization. In this paper we have argued that the moral principles of autonomy, beneficence and justice support reducing SMA through ex-ante methods, like PGD, prenatal testing and termination.
A couple using PGD to create a healthy child confers a great benefit to their next child, when their next child would otherwise have had SMA. This is far greater than the expected benefit generated from ex-post treatments, which are unlikely to cure SMA completely.
When couples engage in prenatal testing and termination and are likely to have another child, this is also a highly beneficial option.
In cases where termination is unlikely to lead to another child, such approaches are not supported by considerations of beneficence.
There are also reasons of justice to promote ex-ante methods.
PGD and prenatal testing and termination are many times cheaper than ex-post treatments for SMA. By prioritizing these more cost-effective therapies, we ensure that medical resources are more equitably distributed.
The advantages of ex-ante interventions over ex-post methods suggest that we should also be prioritizing carrier screening for SMA. It is only when prospective parents know that they are SMA carriers that they can make fully informed decisions about which treatments to employ.
There are therefore reasons of autonomy to promote carrier screening.
Furthermore, insofar as carrier screening will make prenatal testing and PGD more visible options for couples at risk of having children with SMA, such programs will also promote beneficence and autonomy. This is not to say that this is the only way the moral principles could be applied in this case, or that there are no other moral considerations that could bear on these issues. But, in the absence of counter-arguments, we believe that we have provided a strong ethical argument in support of prioritizing ex-ante over ex-post methods to reduce the burden of SMA.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.