Germline genome editing versus preimplantation genetic diagnosis: Is there a case in favour of germline interventions?

Abstract CRISPR is widely considered to be a disruptive technology. However, when it comes to the most controversial topic, germline genome editing (GGE), there is no consensus on whether this technology has any substantial advantages over existing procedures such as embryo selection after in vitro fertilization (IVF) and preimplantation genetic diagnosis (PGD). Answering this question, however, is crucial for evaluating whether the pursuit of further research and development on GGE is justified. This paper explores the question from both a clinical and a moral viewpoint, namely whether GGE has any advantages over existing technologies of selective reproduction and whether GGE could complement or even replace them. In a first step, I review an argument of extended applicability. The paper confirms that there are some scenarios in which only germline intervention allows couples to have (biologically related) healthy offspring, because selection will not avoid disease. In a second step, I examine possible moral arguments in favour of genetic modification, namely that GGE could save some embryos and that GGE would provide certain benefits for a future person that PGD does not. Both arguments for GGE have limitations. With regard to the extended applicability of GGE, however, a weak case in favour of GGE should still be made.


| INTRODUC TI ON
editing into medical practice. 12 The issue of genetic enhancement is also being discussed, but is far from becoming a reality. 13 Conversely, other voices point to the unprecedented risks of introducing irreversible mutations into the human genome. Because genetic modifications can be passed on to subsequent generations, it is hard to contain possible side-effects. Apart from safety concerns, further arguments against GGE call attention to the lack of consent from future offspring, possible eugenic or slippery slope effects, possible resulting social inequalities, and concerns about 'de- Science, ethics, and governance' from the U.S. National Academies of Sciences, Engineering, and Medicine (NASEM) suggested that clinical research and application using GGE should be permitted, and that in some scenarios GGE 'would provide the only or the most acceptable option for parents who desire to have genetically related children'. 16 This conclusion was considered a paradigm shift towards the acceptability of GGE. 17 Despite expressing a favourable opinion, the NASEM do not embrace GGE unconditionally. Clinical trials (as well as future application) are deemed to be permissible only under certain conditions, for example if 'data on risks and potential health benefits of the procedures' are available and interventions are restricted to 'preventing a serious disease'. First and foremost, the NASEM stress that future trials are only legitimate in 'the absence of reasonable alternatives'. 18 A similar constraint has been discussed by other institutions such as the Nuffield Council. 19 Considering the uncertainties and possible RANISCH risks associated with GGE, it is imperative to employ less risky technologies if they can achieve the same ends. Evitt  This paper will explore whether and under which conditions GGE could have any advantage over existing technologies. The focus will be on a comparison between GGE as a type of genetic intervention or modification and embryo selection after preimplantation genetic diagnosis (PGD). This technology of selective reproduction has been widely discussed as a major alternative to germline modification.
However, opinions diverge significantly over possible advantages and disadvantages. Regarding the possibility of embryo selection, some authors see no real benefit of GGE over existing methods such as PGD and thus consider genetic interventions as superfluous. 21 Others point to the limitations of PGD and highlight that these could be overcome in the future by GGE. 22 In addition, some consider GGE to be the morally better strategy, 23 some see it as morally equal to PGD and embryo selection, 24 while others reject GGE in favour of selection. 25

| WHAT ARE THE ALTERNATIVE S?
The demand for GGE is most likely to arise in situations where one or both reproducers are a known carrier or sufferer of a genetic disease and want to prevent the transmission of the disease-causing mutation to their offspring. Thus, the desire to have the chance of conceiving a healthy child drives the development and possible application of this new technique.
It follows that, regarding the permissibility of GGE, it must be asked what can count as a reasonable alternative for intended parents. Arguing that some couples should simply refrain from reproduction would not qualify as such, because this option precludes the desired end, namely starting a family. It must rather be asked what means allow intended parents to have healthy offspring. In this context, the attributed value of having biologically related offspring is a decisive factor. After all, sperm or egg donation, surrogacy, and adoption could allow couples to have a healthy child. But the preference for genetic over non-genetic parenthood is widespread. 26 While there might be pragmatic and perhaps moral reasons to change this preference (e.g. in relation to the possible benefits for orphans), the question of the (il)legitimate moral weight of the desire to have a genetic link to one's offspring will not be discussed in this paper. Rather, the common wish for biological relatedness will be taken for granted here, and the focus will lie on existing and future technological means that could help couples to have biological children.

| Somatic gene therapy
In the context of a disease-carrying couple, two alternatives to GGE are frequently proposed: therapy after birth, and selective repro- However, somatic gene therapy would not be an efficient alternative to GGE in all cases. Even though new therapies could, in principle, be used to correct the specific somatic cells of a newborn or a child and ameliorate the condition, some congenital or early-onset diseases would affect a subject severely before any therapy was feasible. In some cases, for example forms of lysosomal storage disor-   DMD and other disorders also affect widespread and different types of tissues, making it difficult for a somatic therapy to reach all affected cells. 31 Moreover, tissues in which the genetic disease is manifest are sometimes hard to access, for example in neurodegenerative disorders such as Huntington's disease. 32 When a couple is a known carrier of such mutations, GGE on an early embryo could be advantageous over somatic gene therapy after birth, because only single cells of gametes or zygotes would need to be targeted in vitro.
Compared with this, for somatic gene therapy to be effective, a large number of cells would need to be targeted. Furthermore, in the context of GGE, possible therapeutic failures can be contained better, because embryo selection or even abortion remains an option. 33 In addition, successful GGE would have a multi-generational advantage over somatic therapies. Even though GGE cannot eradicate inherited diseases for good, because offspring may develop new mutations or mate with a carrier, GGE could reduce the frequency of mutations in future generations. In summary, while somatic gene therapy could become an alternative to GGE for specific pathologies, direct intervention into the human germline would likely be the more effective strategy in some cases.

| Selective reproduction
Selective reproduction is commonly proposed as another alternative means that allows couples to have healthy, biological offspring.
Selective reproduction encompasses various attempts 'to create one possible future child rather than a different possible future child '. 34 This includes invasive or non-invasive procedures of prenatal testing, which could lead to the selective termination of the pregnancy.
Because abortion is most often invasive and stressful for women, and the moral status of a fetal life is widely considered to be higher than that of the human embryo, selection after preimplantation genetic diagnosis (PGD) must be seen as the preferable alternative to abortion.
Just like most scenarios for germline therapies, PGD presupposes assisted reproduction (e.g. IVF or intracytoplasmic sperm injection). Preimplantation embryos are then analysed for genetic mutations, and only unaffected embryo(s) are transferred. Although assisted reproduction is considered to be safe, hormone stimulation, egg retrieval procedures, and low success rates often put physical, mental and financial burdens on the woman.
Within certain limitations, PGD is permitted under various regulatory regimes. 35 In the U.K., for example, under the regulations of the Human Fertilisation and Embryology Authority, PGD is allowed for more than 500 conditions. In addition to the main application, namely the avoidance of single-gene disorders such as cystic fibrosis, PGD and embryo selection have been used to avoid chromosomal aberrations, to reduce genetic risks (e.g. for breast cancer), for sex selection, and for HLA typing. Some companies claim to be able to screen for some polygenic conditions, 36 and, controversially, fertility clinics began offering PGD to select for cosmetic traits such as the eye colour of the future child. 37 Owing to its wide applicability and its favourable risk profile, PGD is frequently considered to be the major alternative to GGE. 38 If PGD and embryo selection could be used to achieve the same end as direct modification, there seems to be little or no justification for further research and for the development of GGE techniques for human reproduction. Thus, GGE and PGD need to be compared regarding the possible advantages of germline intervention over selection. With this in mind, two related topics will be analysed: the possible clinical (Section 3) and moral (Section 4) advantages of GGE over PGD.

| THE LIMITS OF P G D AND THE CLINI C AL ADVANTAG E OF G G E
While PGD can sometimes give intended parents the chance to have healthy offspring, it is not an effective strategy in all cases. 39 There are scenarios in which PGD will always be useless or where the chances are significantly low that selective reproduction can help intended parents to have a child that does not carry the mutation. In addition, the transfer of unaffected embryos may be feasible, but couples might object to the means or ends of selective reproduction.
The clearest cases where PGD is futile are occasions where a would-be parent is homozygous for an autosomal-dominant disease (e.g. Huntington´s disease or Marfan syndrome) or where both parents are homozygous for an autosomal-recessive disease (e.g. cystic fibrosis). In such cases, it is impossible not to pass on a mutated allele to any future offspring. The same is true for a parent that has a chromosomal aberration in germline cells due to homologous Robertsonian translocation (e.g. leading to Translocation Down syndrome in offspring).
The case of inherited mitochondrial diseases is special. These are often severe diseases with a high variability in symptoms that can be caused by mutations in the maternal mitochondrial DNA (mtDNA).
Although PGD has been used to prevent the transmission of mito- Even if the exact numbers of such constellations were so low as to make these cases insignificant compared with other diseases, it is hard to see how this fact alone could make GGE illegitimate and thereby deny some couples the chance of having healthy offspring.
The argument of rarity does not provide a reasonable moral justification to ban GGE. On the contrary, when direct intervention into the germline is the only or most reasonable option to have healthy offspring, GGE could increase reproductive options for some couples and thus extend reproductive autonomy.
In addition to scenarios where PGD never gives parents the chance to have healthy offspring, more frequently there are cases where it is unlikely (to various degrees) to be possible to select embryos that would not have the deleterious mutation. When both parents are heterozygous for autosomal-dominant conditions, on average three out of four embryos will be affected by the condition. In principle, PGD is an option here, but the chances of conceiving a healthy child are low because the number of unaffected embryos is highly reduced. Here more scenarios for application are conceivable, for example Y-linked gonosomal conditions. 47 In all these cases, a sufficient number of embryos would usually be needed to allow the transfer of suitable embryos. This, however, is a key challenge in assisted reproduction and PGD. Gyngell and colleagues calculated that, in the U.K. alone, more than 120 IVF   Finally, one could imagine couples who could have healthy offspring by using assisted reproduction but who reject selective reproduction on religious or moral grounds. For the first live birth following MRT, it was reported that the woman was motivated by religious reasons to undergo the highly experimental procedure of spindle nuclear transfer. 59 In a similar way, GGE might seem an attractive option for some intended parents because it could avoid the destruction of embryos after PGD and offer a direct fix for affected embryos.
But, as will be discussed in the next section, such a scenario seems highly unlikely.
In view of all the above, the widely held claim that there is no clinical advantage of GGE 'over existing and developing methods' 60 does not hold true. Even though embryo selection after PGD often allows monogenetic disorders to be avoided, it is not a feasible strategy in all constellations. Sometimes direct modification could provide the only possible way to give intended parents the chance to have healthy biologically related offspring. Notably, this conclusion about the extended applicability of germline editing is a factual not a moral statement. The normative implications depend on additional considerations. For now, however, it is safe to conclude that a prima facie case in favour of GGE can be made. Following the recommendations from the NASEM and others, in consideration of the lack of alternatives, GGE could be a legitimate option for certain cases.

| THE MOR AL ADVANTAG E OF G G E
Apart from the extended applicability of germline modification, several sources have proposed that GGE has moral advantages over

| The argument of embryo protection
It is widely believed that human life in its early stage has some value, which constitutes enough reason for embryo protection. This claim is sometimes stated in absolute terms, namely that embryos have the same moral status as adult human beings or persons. More frequently, a moderate version is defended, arguing that human em-

RANISCH
mutations that could be targeted for genetic modification. Then, however, it is not clear why the available and unaffected embryos should not be transferred in the first place. 64 It would be paradoxical to reject suitable embryos after PGD in order to give embryos that carry a disease-causing mutation a chance of being cured. Only after failed attempts to transfer unaffected embryos would it seem plausible to edit otherwise unsuitable embryos that are now routinely discarded. 65 An additional technical hurdle, which could lead to embryo loss, arises here. Gene editing of the embryo should happen early, ideally before the first cell division, that is, directly with or right after fertilization. 66 At this point of development, however, PGD is not feasible without destroying the embryo. Thus, when not all embryos are affected, one either has to neglect embryo testing and apply GGE 'blindly' at an early stage, thereby risking harming some otherwise suitable non-mutant embryos, or applying GGE later after mutant embryos were identied and thereby accepting a loss in efficacy owing to an increased risk of mosaicism. Both options create additional risks of embryos being damaged.
Saving embryos is conceivable mainly in those rare instances described above, where it is expected that all embryos from a couple would have disease alleles. But even then, rather than avoiding PGD, GGE would create an additional indication for testing embryos.
Owing to risks such as off-target effects, success in GGE outcome would need to be validated. Even though PGD does not guarantee to detect off-target effects, in a likely scenario it will still be performed after intervention, to reduce the risk of adverse effects of GGE.
Hence, as long as GGE is not perfectly accurate, embryos might again be discarded even after the intervention.
In consequence, it is unlikely that GGE will have a significant effect on rescuing embryos. It is even less likely that GGE will soon become a 'replacement for PGD'. 67 In a possible future scenario of GGE, embryo-testing and possibly selection will likely be conducted once or even twice: after the intervention and most often before intervention. Thereby embryos might be rejected for transfer either because enough unaffected embryos are available for transfer, or because genetic testing shows that GGE was not successful.
Those who espouse the moral status of embryos can claim a sec- From the perspective of embryo protection, it should be noted, however, that basic research with gene editing will often lead to the destruction of embryos. Most research on GGE so far has used triploid (3PN) human embryos, which are believed to be non-viable, 73 in order to avoid this moral concern. 3PN embryos, as well as orphan embryos, however, are unsuitable research subjects with which to investigate embryonic development. Thus, research on germline editing itself will likely lead to the creation and destruction of embryos.
It has been suggested that this could be justified if genome editing research is likely to reduce global embryo loss in the long run. 74

| The argument of benefit
Irrespective of the possible protection of embryos, it could be argued that GGE has a real advantage over PGD, because direct interventions might benefit a future person in a way that embryo selection does not. This line of argument has been proposed by Gyngell and colleagues as well as other authors: 75 64 Other things being equal, it is plausible to assume that it would be equally good for Ben to be well and alive as it would be for cured-Ana. Then, however, it is not clear what the alleged advantage of GGE amounts to, assuming that the omission of GGE leads to the birth of healthy Ben.
To save the argument of therapeutic benefit from this objection, a particular view on values must be maintained: an outcome can only be better (or worse) for a particular person. This claim can be described as the person-affecting view. According to this, it cannot be argued that a future person is made better or worse off by an action if the same person's existence depends on this decision under scrutiny. From this perspective, PGD and selection were not better for Ben in the second case, because otherwise he would not exist. And, conversely, even if sick-Ana was conceived in this case, she could not complain about her parents' choice to reject PGD, because otherwise Ana would not exist. 80 This stands in contrast to the first case, where we can imagine sick-Ana having a legitimate claim against her parents for not being treated with the pre-emptive therapy.
While the person-affecting view seems to follow naturally from a commonsense concept of 'better' or 'worse', another perspective on such valuations is put forward, too. Defenders of an impersonal view contend that an outcome can be better (or worse) without being better (or worse) for a particular person. 81 Considering the first case, both views come to the same conclusion albeit for different reasons: it is better if the couple decides to use GGE than to forego therapy.
According to the impersonal view, GGE leads to a better outcome, for example more health or wellbeing in the world. From the personaffecting view, GGE is the better option, because it is simply better for Ana to be born healthy rather than sick.
In the second case, the two views lead to different conclusions. Most defenders of personalism try to mitigate these implications. While they maintain that identity-affecting decisions usually cannot be worse for a future person, they grant an exception: 'cases where the child's life is so awful that we can actually deem it worse than non-existence'. 84 In situations of a so-called wrongful life it would then be worse if a child was born. According to a different version of this argument, the presumably wrongful life would constitute harm for this child, while it is otherwise impossible to harm by creating life.
This assumption has implications for the evaluation of the second case, where the choice is between PGD (i.e. healthy Ben) and natural conception (i.e. sick-Ana). If Ana's genetic disease were so severe as to make her life worse than non-existence, it would be better not to conceive the embryo from which she developed.
Accordingly, choosing PGD would be better in the second scenario, too. This claim seemingly makes it possible to uphold some form of personalism, while not being indifferent to the welfare of future people. However, this deviation from the person-affecting view is ad hoc. 85 If an outcome can only be better (or worse) if it is better (or worse) for a person, then even in cases of a horrible life, it cannot be argued that it was better not to exist. Moral limits on reproductive decisions cannot be embraced by the person-affecting view in this way.
There is a second line of defence, which combines person-affecting considerations with impersonalism. According to this view, impersonal concerns matter but to a lesser degree than person-affecting concerns. 86 Hence, in the second case it would be worse to conceive sick-Ana than healthy Ben, but conceiving sick-Ana in the second case is not as bad as failing to cure Ana in the first case. Even if the outcome is the same, a difference is stated here: while in the second case only a worse state of affairs would have been caused, in the first case a worse state of affairs has been brought about and Ana was deprived of being healthy. This claim does not fall into the view that the welfare of certain future people is morally indifferent, while upholding that GGE is in some sense superior to selection.
Such a line of defence, however, suffers from a serious weakness: the proposed benefit of GGE does not only seem to surpass embryo selection. Bringing about a child that has been treated with GGE now seems even more beneficial than bringing about a healthy child. Because then, not only a good state of affairs would be brought about (birth of a healthy child), but also a particular person would have been made better off by being healthy rather than sick. Hence, in the case of a couple that could conceive a healthy child naturally, it would be better to transfer mutant embryos after they were cured. This implausible implication points to a widely held view on reproductive decisions, or rather on the question of benefit in this context, that is just as paradoxical: while we have reasons to prevent lives from coming into existence, because they would be miserable, we seemingly have no reasons to bring lives into existence, just because they would be happy. 87 In other words, creating a child that has a good life does not seem to be any better in moral terms than not creating a child. This notorious asymmetry is a widely discussed puzzle for any ethics of procreation. However, it poses a special challenge for defenders of the argument of benefit. If it is maintained that GGE is better for a future person compared with PGD it needs to be explained, in what sense bringing a healthy child into existence is beneficial in the first place and, why this is even better than selecting a healthy child.

| CON CLUS ION
New tools for genome editing such as CRISPR are widely considered to be disruptive technologies. 88 But when it comes to the most controversial application, the modification of the human germline, there is no consensus on whether GGE has any real advantages over existing procedures such as embryo selection after PGD. The find an answer to this question is crucial for evaluating whether research, development, and the future application of GGE is legitimate. Considering the risks and uncertainties associated with this new procedure, it has been proposed that GGE is only legitimate when no established alternatives are available that could achieve the same ends. This paper has investigated possible alternatives to germline modification as well as the (dis)advantages of GGE over existing reproductive technologies. Considering the extended applicability, it was shown that there are scenarios where GGE would provide the only option for intended parents to have healthy, biologically related offspring. Even though these constellations are comparably small in terms of numbers, they cannot be ignored. Following the recommendations from the NASEM and others, research and development of GGE should then be an option in these cases.
With regard to the argument of embryo protection, it was indicated that GGE has no significant moral advantages over embryo selection. Rather than replacing PGD, in most cases GGE would most likely be a supplemental tool and routinely create additional reasons for genetic testing and embryo selection. The argument of benefit and its underlying view on values was revealed as questionable. If accepted, it leads to the conclusion that the welfare of some future people is morally negligible. A possible modification of this view was not convincing either: it privileges GGE unduly, rendering pre-emptive therapy preferable to the natural conception of a healthy child. 83