Time to start intervening in the human germline? A utilitarian perspective
Focusing on present‐day possibilities raised by existing technology, I consider the normative aspects of genetically modifying the human germline from a utilitarian standpoint. With reference to a hypothetical case, I examine the probable consequences of permitting a well‐conceived attempt to correct a disease‐associated gene in the human germline using current CRISPR gene‐editing technology. I consider inter alia the likely effects on utility of creating healthy new lives, of discouraging adoption, and of kickstarting a revolution in human germline genetic modification (HGGM). I reject various objections to HGGM, including claims that the risks of genetic harm outweigh the likely benefits. From this utilitarian analysis, I conclude that strong grounds exist to support intervening in the human germline using current technology. Delay in commencing such work will impose a utility cost, because the longer we wait until commencing the HGGM revolution and moving towards a world of increased utility, the greater will be the quantity of suffering accrued meantime through genetically influenced disease. Nevertheless, considering residual safety concerns and the negative publicity engendered by an ethically problematic recent (2018) first attempt at HGGM, it seems prudent—and ultimately generative of the greatest amount of utility—to delay implementing HGGM for a modest period of time, in the order of 1–2 years.
Since the 1970s, scientists have been able to genetically engineer the germline of mammals.1 However, in most cases the modifications have involved incorporation of genetic sequences into essentially random sites within the genome.2 Gene targeting was only attainable at very low efficiencies, utilizing ‘cloning’ technology that could not be applied to humans. All this has changed in the last few years, as new genetic modification (GM) technologies have rapidly emerged. It is now possible to reliably perform targeted genomic alterations in somatic and germline cells across a broad range of mammals.3 This technology could be used to intervene in the human germline, a prospect that presents major bioethical issues.
The new GM technologies are based on biomolecules that have been adapted such that they can efficiently locate a specified genomic site and genetically modify that locus. Several such ‘designer’ biomolecules now exist, including inter alia ZFNs,4 TALENs,5 and CRISPR,6 and it is no exaggeration to state that these agents have brought about a revolution in GM.7 At present, CRISPR biomolecules are showing the greatest potential, largely due to the relative ease with which their sequence‐recognition component can be programmed to target specific sites within the genome.8
Very recently (November 2018), scientist He Jiankui, at the Southern University of Science and Technology of China, announced the creation of the first GM babies.9 This work was apparently conducted in secret and, at the time of writing, has not been published in a peer‐reviewed journal. Via a video posted on YouTube, He claims to have impregnated a woman with CRISPR gene‐edited embryos, resulting in the birth of healthy twin girls. The modification entails knock‐out of a genetic pathway exploited by HIV to infect cells.10 At a conference, He subsequently claimed that there is another pregnancy.11 The Chinese government has recently confirmed the births.12
GM technology in general has generated a wide array of ethical concerns, and human germline genetic modification (HGGM) has engendered the greatest disquiet.13 It was no surprise that news of He’s experiment immediately led to a vociferous outcry of moral condemnation from media, society and many academics. More deliberatively, bioethicists have applied a broad range of normative approaches to various aspects and forms of GM. Utilitarianism is one of the most powerful of these approaches; this paper will evaluate the ethics of HGGM from a utilitarian perspective.
Utilitarianism is predicated upon the principle that ethical judgements should be based on the consequences arising, or likely to arise, from a course of action. Desirable consequences are those in which positive mental states (happiness, or ‘utility’) outweigh negative mental states (suffering, or ‘disutility’), in aggregate across all affected individuals.14
While it is not a flawless moral theory, utilitarianism possesses key features that render it valuable, including a high degree of internal consistency and a rational, analytical approach. Importantly, because happiness and suffering (in all their forms) are, respectively, highly valued and strongly deprecated by virtually all agents, conclusions generated by utilitarian analysis arguably have the potential for broader acceptance than those derived using theories based on nonconsequentialist principles or religious precepts. Utilitarianism is particularly powerful as a tool for deciding whether the risk‐benefit ratio of a proposed course of action is favourable—a central issue in much clinical and research ethics work.
While some ethicists (including the present author) are committed utilitarians, most are not. For example, Kantians, Christian ethicists and virtue ethicists work from precepts that are wholly different from, and largely incommensurate with, utilitarian principles. Indeed, some ethicists are strongly opposed to utilitarianism. Inevitably, many non‐utilitarian moral philosophers will disagree with the modus operandi of this article. However, most bioethicists are neither utilitarians nor pure deontologists: many subscribe to some form of value pluralism, or sidestep foundational questions by relying on principlism15 or casuistry16 . While such non‐utilitarian bioethicists do not consider utility as decisive, most do consider it to be a relevant factor in ethical deliberations. Moreover, utilitarian concepts are often borrowed by non‐utilitarian ethicists, such as principlists and some Kantians.17 Accordingly, this paper is of potential interest to a broad audience, including non‐utilitarian bioethicists. I suggest that it should be read by non‐utilitarians as offering pro tanto utility‐based arguments concerning the ethics of HGGM, which can be weighed against fundamental questions of ethical theory, consideration of which lies beyond the scope of this paper.
Several variants of utilitarianism exist.18 Some forms seek to maximize utility through indirect means; for example, through maximization of preference satisfaction, or by establishing ethical rules to be followed in lieu of case‐by‐case utility calculation. These variants, while being valuable in certain contexts, will not be used in the present paper. Instead, I shall employ a hedonistic utilitarian approach, focusing directly on the metal states of individuals.19 Such utilitarian deliberation entails cost‐benefit analysis based upon an approximate quantification of utility change, and this will be the approach taken in this paper in considering the ethics of proceeding with HGGM.
3 HGGM: PROSPECTS AND LIMITATIONS
Once the technology is fully developed, the prospects for HGGM are substantial. Where the genome of one or both of a pair of prospective parents contains a known genetic sequence that causes a specific disorder, a bespoke GM strategy could permanently correct the disorder‐associated sequence in an embryo conceived from these parents. In principle, this would offer a tantalizing means for parents to eliminate or greatly reduce the risk of their child developing the disease in later life.20 It would also avoid the disorder‐associated sequence being transmitted to any future descendants of the GM child.
Many common disorders, such as neuropsychiatric disorders (e.g. schizophrenia, autism), are polygenic in nature, meaning that several disparate genetic loci are involved, each serving as a risk factor. Thus, a future goal of HGGM would be the ability to simultaneously modify multiple genomic sites in an individual embryo.
The human germline per se is by no means ‘perfect’, with evolution having furnished us with rather minimal protection from diseases that tend to strike after the reproductive years. Humans in general are thus at risk from an array of genetically mediated medical problems including cancers and various degenerative disorders. In the future it will likely become possible to modify the human genome to protect against some of these common disorders. This has previously been achieved to an extent in GM experiments on non‐human animals (mainly mice).21 If several common disorders could be avoided or delayed by HGGM, the average disease‐free lifespan could be substantially extended.22
Through disease‐avoidance and disease‐reduction, HGGM promises to boost utility, and is thus a prima facie candidate for utilitarian support. Increased utility can be envisaged at two levels, individual and population. At the former level, HGGM would avoid suffering by the GM individual (and their loved ones) had the intervention not occurred. Regarding the population level, it is apposite to imagine two possible worlds: World A, with no HGGM, and World B, in which HGGM is used. World A will contain more disease and debility, and thus lower overall utility, compared with World B. Utilitarians thus have good reason to support HGGM in principle, as a means to reducing future suffering. However, as with any new biotechnological innovation, questions of safety exist: if a GM approach inflicted genetic damage that led to increased suffering, this would be a disutility, counting against it. More fundamentally, all currently plausible forms of HGGM would require IVF procedures, which would be costly and burdensome for prospective parents. And the availability of HGGM may have the unintended consequence of discouraging the adoption of children. The disutility arising from these negatives must be factored into the utilitarian calculus.
New GM technologies have moved us closer to the above prospects for HGGM. Applied to animal embryos, CRISPR technology can reliably induce highly specific genomic alterations, and it can be used to perform multiple locus alterations in a single embryo.23 Efficiency rates for CRISPR are already high and are continuing to improve, and very low rates of unintended mutations are now regularly achieved.24 Nevertheless, the technology is not yet at a stage where it could reliably deliver many of the more ambitious goals above. This is compounded by an additional limitation: our scientific understanding of the links between genetic sequences and health remains at an elementary stage. While we have detailed genetic knowledge of many monogenic disorders, such as cystic fibrosis (CF), Duchenne muscular dystrophy (DMD) and Huntington’s disease (HD), we know far less about the genetic aetiology of polygenic disorders, such as cardiovascular disease, diabetes, and neuropsychiatric disorders. Therefore, many of the GM applications outlined above remain purely theoretical at present.
Looking further into the future, GM technology will likely become highly efficacious. This promises several utility‐promoting benefits, including correction of the random mutational damage that naturally occurs in everyone’s germline cells, and reversal of previous deliberate genetic modification should this ever be required. It might also become possible to genetically modify sperm cells. Although genetic engineering of sperm is inherently limited to modifying only the paternal contribution to the child’s genome, a sperm‐mediated approach in combination with artificial insemination (AI) would make intervening in the germline much easier (and cheaper) than embryo‐based approaches. This would permit widespread use of HGGM, as opposed to the technology being available only to affluent nations or individuals.25
Beyond modifications to prevent disease, future GM technology could also be used to enhance normal human functioning. Enhancements might include improvements to abilities such as eyesight, physical strength or endurance. The brain might also be enhanced, such as to boost cognitive ability (this has already been achieved in genetically modified mice26 ), or—most controversially—to alter morally relevant behavioural tendencies.27
4 THE IMPORTANCE OF EARLY USE OF TECHNOLOGIES
It is valid, and necessary, to discuss the ethical implications arising from highly futuristic HGGM applications, such as the creation of humans with enhanced abilities, because it seems inevitable that the technology will, one day, become good enough to reliably alter the germline such that these scenarios are no longer science fiction. However, this paper will put distant future possibilities aside and focus instead on the technology as it currently exists, in order to address the ethics of applying available GM techniques to humans. This early‐use possibility is of prime ethical interest per se; and it is also important in that the initial attempts at modifying the human genome will serve as the crucial first steps towards more advanced applications.28
The importance of ‘first steps’ can be illustrated with the paradigm case of IVF. Hitherto restricted to animals, in the 1970s IVF was attempted with a human couple, leading to the birth of a healthy child. This bold step—which was conducted in secret and largely avoided prior ethical scrutiny—kickstarted the revolution in IVF that has transformed the treatment of infertility.29 Had risk‐averse precaution been enacted at the time, preventing the first‐use human IVF attempt, the development of this valuable reproductive technology would likely have been delayed substantially.
However, by citing the IVF case I am not arguing for a simple laissez‐faire approach towards HGGM. The first IVF attempt can be criticized on various ethical grounds. And a lack of caution in medicine can lead to serious negative outcomes, as evidenced by medical tragedies such as the epidemic of birth defects caused by the inadequately tested drug thalidomide.30 Moreover, an early attempt to apply a new technology that results in failure or damage to health risks doing the converse of priming a biomedical revolution, producing instead a proscriptive response from society and regulators. If the technology in question is inherently of potential benefit, the chilling effect on its development resulting from a failed early attempt would be an unfortunate and undesirable consequence.
The claimed creation of the first GM humans by He Jiankui, as described above, is apposite here. The possibility exists that He’s experiment will prove to have been safe and successful. But from what is known so far, the portents are not good. This is so for several reasons, including:
- The experiment was designed to create children who would be resistant to infection from a particular strain of HIV; all the children were produced using sperm from fathers infected with HIV, but there was no need to use GM for this purpose, as sperm‐washing techniques are available that reliably remove HIV particles from sperm.
- Independent scientists examining preliminary data have expressed concerns that there appears to be evidence of mosaicism in placental material and an off‐target edit in one embryo sample. These occurrences, if they have affected the GM babies, could mean that the goal of HIV resistance has not been achieved, and they raise safety concerns.
- There are questions around whether institutional ethical approval was obtained for the work.
- It is unclear whether the parents had freely given their informed consent.
The response to He’s experiment from media and society has been heavily negative, and it seems likely that scientific freedom in this area will be curtailed rather than liberated by what appears to have been an ill‐conceived, ethically fraught and not entirely successful experiment. However, the seemingly botched nature of He’s attempt does not imply that HGGM is inherently undesirable. As described above, major benefits to humanity are likely to accrue from the ability to modify the human genome. Realization of these benefits depends upon the development of effective HGGM technologies, and such development cannot come to fruition in the absence of actual attempts to create GM humans. As opposed to simply banning HGGM—as called for by many commentators following He’s experiment—for utilitarians the correct approach to decide whether it is right to proceed at the present time with HGGM is to weigh up the relevant risks and probable benefits in terms of their likely effects on utility.
5 EARLY‐USE POSSIBILITIES
The disorders most amenable to GM technology are monogenic conditions with high penetrance, in which the association between gene and phenotype has been thoroughly established. For such disorders an appropriate GM strategy could be devised based on current technology that could be applied to embryos from parents whose genome(s) harbour the aberrant gene. Because the necessary technology is well established in transgenic animal research, such a strategy should have a high probability of success; and the concomitant avoidance of an otherwise inevitable serious medical condition would yield a high utility. This combination of high magnitude and probability of benefit gives utilitarians prima facie reason to view serious high‐penetrance monogenic conditions as appropriate candidates for early‐use HGGM.
However, well‐characterized monogenic disorders can already be avoided using existing technology, namely pre‐implantation genetic diagnosis (PGD).31 Consider for example sickle cell anaemia (SCA). This severe form of anaemia is associated with serious health problems including pain, swelling, infection, and cardiovascular damage.32 SCA occurs in individuals homozygous for a mutated β‐globin gene. Where a pair of prospective parents are heterozygotes, there is a 1 in 4 probability that any child they produce will be homozygous for the mutant gene, and thus develop SCA. However, PGD can be utilized in such cases: IVF is used to produce embryos, which are grown briefly in vitro until comprising a suitable number of cells, at which point a biopsy is taken for genotype testing. Embryos typed as homozygous (or heterozygous) for the wild‐type β‐globin gene can then be selected for uterine transfer and gestation.
GM could in principle be used to the same end. Experiments with cells have established that it is possible to use GM techniques to correct the mutant β‐globin gene.33 Adapting this to the human embryo, a possible strategy would involve delivery of CRISPR biomolecules designed to target the β‐globin gene, along with a correction template DNA molecule based on a wild‐type β‐globin gene sequence; the template sequence would interact with the genome at the CRISPR‐targeted location, in a process known as homology‐directed repair (HDR), thus overwriting and correcting the genetic aberration.34 Treated embryos would then be grown in vitro and subject to genetic sequencing, to allow selection of embryos in which the desired modification had been achieved.
Various CRISPR‐based strategies are possible for other monogenic disorders, with proof‐of‐concept demonstrated in cell and animal GM studies. For example, DMD, an X‐linked dominant disorder caused by a non‐functional dystrophin gene, might be avoided by using CRISPR to delete a key RNA splicing sequence, allowing a modified functional gene product to be expressed.35 And some dominant disorders might be avoided by using CRISPR to knock‐out an aberrantly expressing gene, such as mutant huntingtin in HD.36
Because PGD can be used to avoid inherited monogenic disorders such as SCA, DMD and HD, it would arguably be ethically problematic to attempt GM as an alternative. Use of the former technology in humans is well established and has no special safety concerns (beyond the risks associated with IVF), while GM approaches are unproven in practice. I will argue later in this paper that the health risks from current GM techniques are likely to be low. However, any additional risk would be difficult to justify, given that present GM technology offers no clear advantages over PGD for these conditions. Nevertheless, utilitarians have good reason to support attempting GM in place of PGD on the basis that it would bring about an important wider benefit, namely the kickstarting of a revolution in HGGM, leading to the development of technologies able to deal with polygenic conditions and provide broad improvements to human health.
Moreover, in some cases PGD is of no use, namely where all conceptuses will contain the disease‐causing sequence. For example, where both prospective parents are homozygous for the mutant β‐globin gene (and thus affected with SCA), no embryo they produce will contain a wild‐type β‐globin gene. PGD is also ineffective where one parent is homozygous for a dominant disease‐causing allele. For example, an individual whose huntingtin genes are both of the mutant type can only give rise to embryos that contain a disease‐causing gene. It is such couples who would be potential candidates for early‐use HGGM. A hypothetical couple will be used later in this paper to discuss the ethical issues around an early use of HGGM technology. Before this, an important alternative option to HGGM will be discussed: adoption.
6 IS ADOPTION THE ETHICALLY PREFERABLE SOLUTION?
Two forms of reproductive technology are available to prospective parents who cannot use PGD: donor insemination and embryo donation. And a third, non‐technological way to prevent transmission of a genetic disorder is also available: adoption. Before discussing the ethical aspects of HGGM per se, I shall consider whether adoption might be ethically preferable. This may seem a digression, but it is not: procreative choice is a crucially important factor in the ethics of reproductive technologies, and adoption is thus highly relevant to the ethics of HGGM.37
In contrast to PGD or HGGM, donor insemination or embryo donation will produce a child who is genetically related to only one of the parents; and adoption entails that neither parent is related to the child. Many couples have a strong desire to have children that are genetically their own. This well‐known and commonplace—although certainly not universal—strong desire is evidenced by the substantial financial and emotional costs borne by many infertile couples who choose to use reproductive technologies to produce a related child. It is therefore to be expected that many prospective parents at risk of transmitting a genetic disorder, for whom PGD is not possible, will prefer HGGM—if they perceive it to be sufficiently effective and safe—over adoption.
The fact that many prospective parents desire genetic relatedness, and use reproductive technologies to secure a related child over adoption, does not amount to a moral argument in favour of relatedness.38 From a utilitarian perspective, we must ask: which is preferable, (a) adoption, or (b) a technological solution? It is well known that, worldwide, many orphans are languishing in institutions or surviving on the street; it may reasonably be assumed that they have lives that are on average of substantially reduced happiness compared with other children. Therefore, if option (a) pertains, a substantial utility gain will occur whenever any of these children become adopted by caring parents with the means to support them. If instead the prospective parents opt for (b), the effect will be to add utility to the world by creating a child who is likely to experience a happy existence.39
It is difficult to assess the relative utility gains likely from (a) the enhancement of an existing life through adoption versus (b) the creation of an additional happy life, and therefore difficult to determine with certainty which option is preferable. More fundamentally, the notion that the creation of extra happy lives can compete with the quality of existing lives is contentious. Utilitarianism seems to entail that it is ethically desirable to increase utility through the creation of happy lives, a position I shall refer to as the utilitarian total view (UTV). But this view is intuitively unappealing, and leads to what Derek Parfit has famously called the repugnant conclusion:
Compared with the existence of many people who would all have some very high quality of life, there is some much larger number of people whose existence would be better, even though these people would all have lives that were barely worth living.40
Avoidance of the repugnant conclusion (RC) is one of the major motivations for rejecting the UTV. However, rejection of this view seems to imply rejection of utilitarianism more generally, as the summation of happiness in aggregate is a central utilitarian principle. Accordingly, some utilitarians have argued that the RC is unavoidable and thus ought to be accepted.41 Such pro‐RC positions are based either on the grounds that our intuitions are outweighed by the strong theoretical basis of utility quantification, or that our intuitions are deceiving us, with the apparently counterintuitive nature of the RC being an illusion.42 However, while these RC‐accepting approaches have the merit of consistency in respect of the principle of utility aggregation, they remain intuitively and philosophically unappealing to most ethicists, including many utilitarians.
The RC may not be a practical ethical issue: in the current world, the creation of additional lives through current reproductive technology would not be of sufficient magnitude to substantially (if at all) reduce the average level of utility.43 Utilitarian bioethicists may thus choose simply to ignore the theoretical implications of adding more lives, sidestepping the RC. While this may be a practical solution, it is intellectually unsatisfying, and this has motivated various arguments in defence of the notion that the addition of happy lives is desirable yet does not entail the RC. I shall call this notion the limited UTV. It is beyond the scope of this paper to discuss the forms of limited UTV that have been proposed by various ethicists; suffice it to say that, despite many such attempts, the RC remains a substantial and largely unresolved philosophical issue in utilitarianism.44 Given this reality, I shall argue from a utilitarian position that does not depend upon the UTV. Nevertheless, I shall refer to this view where relevant to the discussion.
The foregoing considerations about the merits of creating versus improving lives might be considered relevant by individuals making personal moral decisions about founding a family. Certainly, the arguments will be salient for those who consider themselves to be utilitarians. Genetic relatedness is not of itself a source of utility, and there is much to be said in favour of a personal decision to enhance the life of an existing child by adoption, considering the substantial utility increase that is likely to accrue. For utilitarians who reject the UTV, adoption appears to be the morally superior option as, according to this perspective, the happiness gained by rescuing an existing child is factored into the utilitarian calculus, but the utility produced by creating an additional life does not count.
However, it seems probable that most prospective parents who strongly desire a genetically related child are unlikely to be swayed by such arguments. If so, and if we assume that utilitarian reasoning favours adoption over technology (i.e. that the UTV is invalid), it follows that prospective parents who are at risk of transmitting a genetic disorder would need to be disallowed access to HGGM technology (once it becomes available), so that they have no option to have a (healthy) child other than through adoption.45 Such a rule would reduce the parents’ utility, by denying them the ability to satisfy their strong desire to have a genetically related child. But the balance of utility might still be in favour of adoption: it seems reasonable to assume that, in most cases, the utility gained by a child rescued from a very unhappy existence would outweigh the utility loss caused to the parents through being denied the means to produce a genetically related child.
There is however a practical problem with a rule intended to promote adoption by preventing access to reproductive technologies: while some prospective parents may decide to adopt, many others may opt not to have a child at all. And it is plausible to think that many of those who would adopt might have done so regardless of the rule, as in many cases they would be individuals who happened not to possess a strong desire for genetic relatedness. All this would count against the rule, because a society in which utility was lost through thwarting of parental desire, with only a small utility gain from additional adoptions, would be less desirable than a society in which parental desires were satisfied, with a concomitantly higher level of utility.46 Of course, this position is tentative, as we cannot be certain as to how people would actually feel and respond in the face of such a rule.47
I conclude that, while adoption per se is strongly supported on utilitarian grounds, arguments favouring adoption over reproductive technology do not amount to a convincing refutation of the position that in principle HGGM ought to be supported by utilitarians. On this basis, I turn now to consider the hypothetical case of a pair of prospective parents, whom I shall refer to as Couple X: both suffer from SCA, and are potential candidates for HGGM.
7 COUPLE X
That both members of the couple suffer from SCA is a realistic scenario, as the disease‐associated form of the β‐globin gene is present at a high frequency within several populations worldwide.48 Suppose Couple X wish to have a SCA‐free child, and they strongly desire this child to be genetically their own. PGD is not possible, as all their embryos will be homozygous for the mutant gene. Now suppose a group of scientists is willing to make (and fund) a HGGM attempt, of the type described above.49 Would it be ethically acceptable to allow this attempt?
This is a complex question, and I shall answer it in several broad stages: I will first consider whether the attempt is likely to be successful; I will then briefly address nonconsequentialist objections to HGGM; before moving to consider issues of harm, risk, consent and precaution.
As described above, CRISPR could be deployed on embryos from Couple X, with the aim of correcting of one or both mutant β‐globin genes. Mice are the main testbed for HGGM techniques, and recent studies have consistently shown impressive results for the effectiveness of CRISPR‐based germline GM: embryo survival rates of around 95% following delivery of CRISPR biomolecules and template DNA have been obtained, and HDR‐mediated correction rates of over 90% have been achieved in the surviving embryos.50 Such work demonstrates that CRISPR‐based GM can be highly effective.
In 2015, the first CRISPR‐mediated GM attempt on human embryos was published: in a study that was widely reported by the news media, researchers in China reported that 71 from 86 (82.6%) tripronuclear51 embryos survived CRISPR treatment, of which 7 (9.9%) had undergone HDR‐mediated modification of the β‐globin gene.52 Subsequently, a separate group targeting other genomic sites, also in tripronuclear human embryos, reported similar results (13.0% HDR).53 Most recently, a group targeting the myocardial disease hypertrophic cardiomyopathy (HCM) gene have reported HDR‐mediated correction in 13 from 18 human embryos, a success rate of 72.2%.54 Notably, this latest work employed normal (not tripronuclear) human embryos, which may in part explain the higher success rate.55
From the above work, the chances of success of a CRISPR‐mediated germline correction in embryos from Couple X should be high. Would this approach likely produce a gene‐corrected child in practice? Suppose that, for Couple X, superovulation yields 10 eggs (a typical number); from the above findings, we might expect perhaps 8 to survive the necessary IVF and CRISPR procedures. Across the few GM experiments that have been conducted on human embryos, the reported HDR rate varies widely, from around 10% for the initial studies to over 70% for the most recent work; thus it is difficult to predict the success rate for Couple X. However, taking a conservative assumption of 25%, this would give 2 embryos with the genetic correction. The necessary embryo biopsy testing step would not be expected to significantly reduce viability, and ongoing pregnancy rates (after 12 weeks) following embryo testing are typically 50–60% for a transferred embryo. Thus, it is reasonable to anticipate that Couple X will be able to produce a gene‐corrected child within a single ovarian cycle. Repeat attempts would also be possible, and would increase the overall chances of success, albeit with utility costs from the burden of IVF cycles.
If this likelihood of success indeed pertains, what would be the effects on utility? To answer this, it is necessary to compare the levels of utility likely from alternative courses of action. Because PGD is not possible for Couple X (as they are homozygotes), the alternatives will be as follows:
- Use donor gametes.
- Use a donor embryo.
- Adopt a child.
- Conceive a child naturally.
- Do not have a child.
In the case of alternative A, either donor sperm would be used to fertilize the female’s oocyte, or a donor oocyte would be fertilized with the male partner’s sperm and transferred into the female’s reproductive tract. In terms of invasiveness, a GM approach would entail superovulation, oocyte recovery, and embryo transfer. Because a donor gamete approach entails fewer invasive steps, it should be less burdensome for the mother. In this respect, her utility level should be higher if a donor gamete approach is used instead of GM. However, this could not fulfil the strong desire that Couple X have for full genetic relatedness: with AI the child would be related only to the mother, or with a donor oocyte, related only to the father. If, given a free choice, the couple prefer the GM approach, this demonstrates that they place a higher value on relatedness than they place on the reduced burdensomeness of a donor gamete approach. If so, their utility gain would be greater if they created a fully related child via HGGM than if they were denied it and instead reverted to alternative A.56
Alternative B (embryo donation) carries approximately the same degree of invasiveness as A, but would be even more problematic in terms of Couple X’s desire for relatedness, as the child would be related to neither parent. Thus, if the utility of Couple X would be lower using donor gametes versus HGGM, the use of a donor embryo would result in yet lower utility, insofar as the happiness of Couple X is linked to the fulfilment of their desire for a genetically related child. In terms of the utility of the child, the use of any of these techniques—HGGM or the alternatives A or B—should be the same in all cases.57 Thus, the personal utility of Couple X is the factor of key relevance in the utilitarian comparison of these techniques. (And for those who accept the UTV, the addition of an extra life would be desirable regardless of which approach was employed.)
The situation is different for alternative C (adoption). If they adopted, Couple X would be increasing the utility of an existing child. To make a utilitarian comparison of adoption with HGGM, it is necessary to consider both the utility of the parents and that of the child. In the previous section, I suggested that, while adoption per se is strongly supported on utilitarian grounds, the deep desire for genetic relatedness held by many prospective parents implies that society ought to establish an adoption‐enhancing rule preventing access to reproductive technologies. However, if Couple X could be consensually persuaded to choose to adopt instead of using HGGM, this would be more desirable from a utilitarian perspective. (This applies with the greatest force to the extent that UTV arguments are deemed invalid.) However, it seems unlikely that logical persuasion or education of such prospective parents about adoption ethics would generally be effective in dislodging strong emotion‐based motivations to have related children.
If HGGM is unavailable to Couple X, and they place such value on full genetic relatedness that they reject alternatives A–C, their only remaining procreative option is D (natural conception). This would produce a child who will develop SCA. While this child may be expected to have an existence that is on balance one of positive utility, SCA will almost certainly substantially reduce the child’s happiness. By contrast, a child produced using HGGM should (all else being equal) experience an existence of substantially greater happiness. Thus, utilitarian reasoning leads to the conclusion that it is ethically preferable for Couple X to choose the GM approach rather than natural conception. This option would also tend to increase the utility of Couple X, as they will avoid inter alia the emotional anguish that is frequently entailed by parenting a child with a serious disorder. Some would claim that the rigours of coping with an ill child ultimately benefits the psyche, making the parent (and possibly the child) a ‘better person’. This Nietzschean claim that adversity makes us stronger—which is also a common feature of several religiously motivated moral perspectives—may have some validity. But from a utilitarian viewpoint, the justification for choosing to knowingly conceive a seriously ill child instead of a healthy child relies on the suffering of those involved (parents and child) being outweighed by the utility gained through psychological strengthening. This would pertain where an individual’s adversity‐improved character leads them to create much good in society. But while this positive balance of utility may arise in the case of a few individuals—whom we might view as moral saints—it seems unrealistic to imagine that this could apply more generally. I suggest that for most parents and their children the effects of coping with a serious and debilitating lifelong disorder will be a net reduction in their utility, and one that will rarely be balanced by wider utility‐generating benefits.
Suppose Couple X are denied HGGM: if they value genetic relatedness so highly that they decline alternatives A–C, and if they decide not to conceive a child who will develop SCA, then alternative E (do not have a child) will be the default. This outcome will result in the couple experiencing less happiness than had they been able to produce a child through HGGM.58 Therefore, on utilitarian grounds the no‐child option is less desirable than the HGGM option. (And for those who accept the UTV, the non‐creation of an additional (healthy) life is ethically undesirable as it entails a loss of potential utility).
Of course, the above assumptions about success rates might be over‐optimistic. The GM attempt with Couple X might fail to produce a viable gene‐corrected embryo and subsequent child. This would have implications in terms of utility. An IVF cycle is burdensome for the mother, entailing inter alia ovarian hyperstimulation and oocyte recovery. Pregnancy itself can be arduous—and a failure at this stage can impose a particularly heavy cost, in terms of the emotional impact of miscarriage. However, given that PGD is impossible for Couple X, and the only viable disorder‐avoiding alternatives (donor gametes/embryo or adoption) are unpalatable for them, their decision to attempt HGGM could be entirely rational, and compatible with utilitarian reasoning. Assuming they freely consent, having been furnished with an understandable and accurate account of the procedures, likely success rates and uncertainties, their decision to proceed represents a rational bet. If the outcome is failure, utility falls; whereas if the attempt is successful, the couple’s happiness should increase markedly.
Beyond the benefits to Couple X of a successful attempt, the likely effect of kickstarting a HGGM revolution is of substantial societal importance. Once HGGM is in use, it is reasonable to expect the technology to improve, rendering it more reliable, efficacious, and cost‐effective. In such a future world, where advanced GM technology has become widely available, one would expect HGGM to be widely used as a disease risk‐reducing add‐on to conception, as opposed to being employed only in cases of monogenic disorder transmission (such as Couple X). As most prospective parents are not at high risk of transmitting a severe, high‐penetrance disorder, the availability of HGGM should not have a broad impact on procreative decisions; therefore, adoption rates ought not to be substantively reduced. Because such a world would contain a reduced amount of genetically influenced disease, with no substantial impact on adoption choices amongst these prospective parents, its overall level of utility should be higher than in an equivalent non‐GM world.
Thus, because the successful initial use of GM on the human germline is a necessary step in the development of more widely applicable HGGM, I conclude that there exists good reason for going ahead with the sort of GM attempt suggested for Couple X. Initial HGGM attempts would be on a small scale: this is inevitable, given that only a small subset of couples at risk of transmitting a serious genetic condition would be candidates for early HGGM. This means that even if a reduction in net utility arose through reduced adoption arising from HGGM in this limited context, it would likely be outweighed by the utility‐generating effects of bringing forward the development of advanced, widely applicable HGGM technology. By contrast, a delay in starting to use HGGM with humans would impose a utility cost, because the time that passes before HGGM starts to yield benefits is time in which extra suffering through genetically influenced disease will occur.
8 OBJECTIONS TO EARLY‐USE HGGM ATTEMPTS
Any proposal to proceed with an early attempt at HGGM (such as that suggested above) would undoubtedly be met with a tranche of objections. These would include several well‐rehearsed positions and arguments, including claims of unnaturalness, the alleged interests of embryos, questions of identity, fears of eugenics, and simply the ‘yuck factor’.59 Most of these objections have little to do with utility, and as such carry negligible weight for utilitarians. However, where disutility is implied by any of these objections, it must be factored into the utilitarian calculus. One source of disutility would stem from unhappiness experienced by the GM individual regarding their identity, if they became distressed over their unnatural manufacture.60 However, this outcome seems rather implausible. Similar concerns should apply to IVF as an infertility treatment, yet there exists no good evidence that people created through IVF suffer unhappiness through knowing that they were produced through an unnatural process.
Another disutility might be a degree of anxiety or frustration experienced by some bystanders in the face of something happening which they consider as morally outrageous.61 A reduction in utility of this sort will of course be experienced mainly by those who have strong cultural or religious objections to HGGMs. For the utilitarian, the question is whether such disutility would outweigh a gain in utility. A substantive increase in utility will be expected in the parents who choose to attempt HGGM. (Also, bystanders who support HGGM will, to an extent, be made happier by the knowledge of a HGGM attempt.) And in terms of the pump‐priming effect of early‐use HGGM, in a future world in which the technology is used to markedly reduce the incidence of disease, there would be substantively less suffering than in an alternative non‐HGGM world. In sum, the suffering of anti‐HGGM bystanders would be outweighed by increased utility arising from HGGM.
9 GENETIC HARM
Utilitarians must take seriously the risk of genetic harm that may arise from a HGGM attempt involving Couple X. It is of course the GM child who would be at potential risk of any adverse effects from the CRISPR technology. Discounting the general risks from routine IVF (which are minimal), the specific risks here are twofold: mutations and mosaicism.62 Mutations occur when CRISPR biomolecules interact with and alter sequences other than those intended to be modified. This usually involves distal parts of the genome (‘off‐target’ mutations), but can also affect sequences close to the target locus (‘on‐target’ mutations). Mosaicism is the generation of an embryo that comprises cells of differing genetic constitution: some cells genetically modified, and others not. If a mosaic embryo was used to create a person, their body would also comprise a mix of cells. Phenotypically, this might result in a child who goes on to manifest some features of the disorder that was to be avoided by the GM. Additionally, the occurrence of a GM‐induced mutation in combination with mosaicism would render pre‐implantation testing problematic, as there would be a risk that the mutation could be missed through analysing only cells that happened not to contain it.
Mutational effects and mosaicism were found to occur at relatively high levels in the early days of CRISPR experimentation.63 However, the accuracy of the technology has been rapidly improved through extensive work at the in vitro cell and transgenic animal levels, with several recent experiments showing extremely low levels of CRISPR‐induced mutations64 or mosaicism.65 Such progress is also reflected in the abovementioned experiments on human embryos. In the first of these studies, the researchers randomly selected six of their CRISPR‐modified embryos for thorough sequencing: of these, off‐target mutations were evident in two cases.66 By contrast, in the most recent work, all the HDR‐corrected GM human embryos obtained were subject to thorough sequencing and no CRISPR‐associated mutations were found.67
However, a recent prominent study, authored by Kosicki et al., has suggested that mutational damage from CRISPR may arise at a higher rate that hitherto has appeared to be the case.68 This research involved GM of various types of mouse and human cells, and discovered that significant on‐target CRISPR‐induced mutations had occurred, including large deletions and various complex genomic rearrangements. The Kosicki et al. study was published in a reputed journal and the work is of high quality: accordingly, it has attracted substantial attention and generated some commentary expressing doubts about the potential safety of CRISPR for use with humans.69 But the results obtained in the study are of questionable significance in the context of HGGM. The findings are surprising, in that very large mutations are described that should have been detected in previous studies that have used whole genome sequencing or other powerful methods to detect mutations. There are several reasons to suspect that these results may not be representative of CRISPR outcomes in general, as follows. Firstly, the work focuses on a form of editing that depends on non‐homologous end joining (NHEJ); this type of editing is known to be error‐prone, and is of less relevance to HGGM attempts than provenly reliable approaches, such as those using HDR (as described earlier in this paper). Secondly, many of the reported experiments in the study selected for cells that exhibited loss of target gene function, potentially skewing the results towards unusual large‐scale mutational outcomes. Thirdly, the study assayed a narrow range of loci, and these may not be typical of target genes in general. Fourthly, most of the reported experiments used a system for delivering the genome editing components that itself is known to frequently cause genomic rearrangements; such a delivery system would not be appropriate for HGGM. Finally, the cell types used were specialized laboratory strains, which probably have different cellular DNA repair processes than exist in actual human embryos, and there is good reason to expect that some of these cell types will be particularly prone to mutational errors during editing.
While the Kosicki et al. paper should not be ignored, there is good reason to believe that the low or zero occurrences of mutations and clinically relevant mosaicism reported in most of the recent literature would likely apply to a well‐conceived, real‐life attempt at HGGM. It does appear to be the case that mosaicism occurred in the He Jiankui HGGM experiment. However, as described above, He appears to have been operating as a rogue scientist, and his work can hardly be said to have been well conceived. The He case points towards the need for HGGM to only be conducted under well‐regulated environments by competent individuals. Most importantly, any ethically acceptable HGGM attempt would incorporate appropriate genetic testing prior to the generation of a GM child. This would include preliminary testing work carried out using cells from the prospective parents,70 to equilibrate the CRISPR and template molecules and to assay for mutational effects (as was conducted as part of the above‐mentioned 2017 human embryo study). Subsequent to such testing, embryos would be screened using PGD coupled with deep sequencing to reveal any mutational issues, such that mutated embryos can be rejected.
It would in any case be wrong to assume that a mutation in the child’s genome would inevitably have an effect on their health. It is well established that only around 2% of the human genome comprises protein‐coding genes and their linked control elements. This implies that any single random mutation would have an approximate likelihood of 1 in 50 of occurring within one of these crucial regions. However, this figure of ca. 2% does not include various genomic elements that express non‐coding RNA molecules hitherto considered functionless, but for which evidence is accumulating of a role in ‘fine tuning’ gene expression and thus potentially influencing phenotype.71 But theoretical considerations suggest that the functional fraction of the human genome cannot exceed 25%, and is almost certainly considerably lower.72 Recent studies indicate that the non‐coding functional regions account for around 6% of the genome, meaning that (in combination with the 2% above) the total ‘target’ for a potentially damaging mutation is around 8% of the genome.73 This suggests that a single CRISPR‐induced mutation would have only around a 1 in 12 likelihood of occurring in a functionally significant part of the genome.
Even where an individual mutation has occurred in the functionally important 8% of the genome, and has somehow escaped detection such that it is present in the GM child, in many or most cases there will be no adverse phenotypic effects. This is most likely in the case of the proportion of the genome that expresses non‐coding RNA sequences with a functional role. In general, these non‐coding RNAs appear to exert only a modest effect on gene expression and, importantly in the context of mutagenesis, their functionality in general does not depend on the high levels of sequence precision that pertains to the ca. 2% of the genome that comprises protein‐coding and other vital sequences.74 In keeping with this model of low criticality, while genome‐wide association studies (GWAS) have linked these regions with disease, in general they appear to act as risk factors (as opposed to disease determinants), each having a minimal contribution to the disease phenotype.75 Thus, a CRISPR‐induced mutation that alters one of these non‐coding elements is unlikely to lead to the child manifesting a genetic disorder.
Even mutations in coding sequence can be functionally silent; for example, affecting part of a sequence in a way that does not change the amino acid constitution of the gene product, or changing it in an insignificant way. Of mutations that do have a negative effect on the gene product or its expression, not all will lead to a genetic disorder; for example, a mutation that knocks‐out a crucial developmental gene may result in a non‐viable embryo (and thus not give rise to a foetus or child), and some mutations that knock‐out gene function will be compensated for by haplosufficiency.76
From the above scientific considerations, it seems clear that a well‐conceived HGGM attempt, such as that suggested for Couple X, is very unlikely to lead to a genetic disorder in the GM child. Nevertheless, it cannot be said that HGGM using current technology would be absolutely risk‐free. The question is, are the risks sufficiently low such that it is ethically acceptable to proceed with early‐use GM?
10 RISKS IN CONTEXT
We are faced with a chicken‐and‐egg dilemma: the real risk rates for CRISPR‐based HGGM cannot be determined until a number of children have been produced using this technology. Thus, if human GM were deemed permissible only once direct proof of an acceptably low risk is available, then it could never be permitted. While many opponents of human GM would no doubt support this position, utilitarians have reason to think otherwise, considering the potential benefits to humanity that are likely to ultimately flow from HGGM.
It is true that an accidental CRISPR‐induced mutation could lead to a negative outcome for the health of the child. However, natural reproduction itself is by no means free from mutational risks. Genetic sequencing has been used in a number of recent studies to analyse genetic differences between trios of individual parents and their (naturally conceived) offspring, and the results obtained so far indicate that natural de novo mutations affect the germline at a rate of approximately 70 de novo mutations per generation.77 This means that each individual is, on average, expected to harbour around 70 genetic mutations that were not present in either of their parents’ genomes, which have arisen in the gametes or conceptus from which the individual developed. Most of these de novo mutations do not cause disease, for the reasons given above. Nevertheless, the risk of genetic disease is positively correlated with the number of mutations in the genome. In epidemiological terms, ca. 1% of neonates suffer from a clinically important monogenic defect, and a further ca. 2% suffer from a congenital malformation, with a substantial proportion of these problems likely being the result of de novo mutations. And several studies have confirmed that the total number of mutations increases with parental age, with children from older parents (particularly fathers) on average harbouring a greater number of mutations.78
This background of de novo mutations provides some perspective from which to view the risk of mutations from CRISPR‐mediated HGGM affecting the genome of a child from Couple X. Of the many new germline mutations that unavoidably arise in the genome of all humans, the majority do not lead to genetic disorders, and thus each such mutation might reasonably be viewed as, on average, contributing an insignificant risk. On this view, the small additional risk from CRISPR ought also to be viewed as insignificant. However, it does not follow from ‘we tolerate risk X’ to ‘we should tolerate risk X+n, where n is smaller than X’. To think otherwise may amount to a context illusion—namely a failure to acknowledge a small potential source of disutility when it is set in the context of a much larger background of the same.79 It follows that, if the increased risk is to be accepted, it must be worth the anticipated benefits. As discussed above, early HGGM attempts such as that suggested for hypothetical Couple X would be likely to deliver substantive utility‐promoting benefits, including fulfilment of their strong desire to have a genetically related child, and the avoidance of instead creating a non‐GM child who will develop SCA. Although it is possible that this gain of utility might be outweighed by disutility through discouragement of adoption, the broader effect of promoting the development of more advanced GM technologies with wide future applicability amounts to a substantial utility boost. The wider future use of HGGM to reduce the incidence of common genetically influenced diseases would almost certainly outweigh the small additional mutational risk associated with current CRISPR technology. Additionally, it is likely that when the HGGM revolution is underway, future technical refinements and advances will be forthcoming that will further reduce the rate of GM‐induced accidental mutations.
11 CONSENT AND HARM
Another potential objection to the proposed HGGM attempt is that the future child cannot consent to the procedure. In other words, the use of GM would violate the autonomy of the child, because the child must bear the risk of a mutational accident without having a say in the matter. Utilitarians have good reason to respect the principle of autonomy,80 and therefore must take the lack‐of‐consent charge seriously. However, a principled objection to GM on grounds of autonomy would also entail the rejection of all pre‐conception interventions, including IVF and PGD. This conclusion is not viable for utilitarians, as established technologies such as these clearly lead to beneficial outcomes, and there is no evidence that lack of consent for prenatal interventions has a negative effect on the happiness of the child. More fundamentally, it is not conceptually possible for a lack of consent to cause disutility, as prenatal consent is impossible. Children cannot consent to being conceived, nor to their genetic constitution, nor to the societal circumstances into which they will be born. Therefore, a strict principle of consent would entail that procreation in general be considered unethical.81 This anti‐natalist position is manifestly untenable on utilitarian grounds, because the application of such a restrictive principle would clearly reduce utility.82 By contrast, a policy based on proxy consent (given by the parents), will increase utility and thus should be considered ethically appropriate for HGGM, just as it is for established reproductive technologies.
From a utilitarian perspective, the key issue here is not consent; rather, it is whether the risk of harm outweighs the likely benefits to the future child. If Couple X decide that they must have a child (i.e. non‐parenthood is unacceptable to them) and that the child must be fully genetically related to both parents (i.e. they will not accept adoption, donor gametes or a donor embryo), the only family‐founding options are:
A. Undergo CRISPR‐mediated GM.
B. Conceive a child naturally.
Option A will (if a viable modified embryo is obtained, and testing shows that the genetic correction has occurred) produce a child who does not develop SCA, but who is at some (albeit low) risk of sustaining a CRISPR‐induced unintended mutation, which might impact on the child’s health. By contrast, option B will produce a child who is virtually certainly to develop SCA. Confronted with only these two options, utilitarian reasoning leads to the conclusion that, because outcome A is preferable to outcome B, it is ethically preferable for Couple X to choose to undergo CRISPR‐mediated GM.
If Couple X choose to knowingly conceive a child who will develop a serious disorder when they could have produced a child who will not develop the disorder, have they harmed the child? Most utilitarians accept the notion of ‘impersonal harm’, in the sense that a procreative choice that knowingly would create a child who will experiences a life of reduced happiness is an act that causes harm, despite no existing person being harmed when the procreative decision was made.83 If a procreative decision is likely to produce a child whose life would not be worth living (i.e. containing more suffering than happiness), then utilitarians have good reason to view such a child as having been harmed, and indeed wronged, by the decision to proceed. On this view, non‐existence (i.e. the no‐child option) would be preferable to such a life. Suppose, instead, that a procreative decision will produce a child whose life would be of reduced happiness but would contain more happiness than suffering. Has harm occurred? If an alternative procreative choice was available that would instead have produced a child who would be expected not to have a life of reduced happiness, it follows that (impersonal) harm has indeed resulted. Accordingly, from a utilitarian perspective it would be wrong to choose to create a child of reduced happiness where an alternative was available.84
In the case of Couple X, a decision to conceive naturally would, by creating a child with SCA, result either in a life not worth living or—more likely in the particular case of SCA—a life of substantially reduced happiness. By contrast, HGGM would avoid SCA while imposing a small risk of producing a child with a CRISP‐induced genetic disorder. In this specific scenario, i.e. where other reproductive options have been ruled out, HGGM would be the ethically preferable choice.
As discussed above, on the assumption that the UTV is invalid, it might be preferable if those in the same (or equivalent) genetic position as Couple X were to choose adoption instead of HGGM, as this would most likely generate the greatest amount of utility. However, to the extent that some of these couples would decline adoption due to possessing a strong desire for relatedness, enabling access to HGGM would—despite the (likely very small) risks entailed by current GM technology—result in increased utility through a reduction in the level of harm that would otherwise accrue when some of these couples chose to conceive naturally. While making the technology available would likely deter some couples from choosing adoption, it would also have the effect of kickstarting the HGGM revolution. This would move us closer to a world in which overall utility would be substantially elevated due to reduced levels of genetically influenced disease.
12 PRECAUTIONARY PRINCIPLE
As discussed above, while the likelihood of harm appears to be low for a first attempt at HGGM using current technology, the possibility of catastrophic mutational accidents cannot be ruled out, although it is impossible to accurately quantify the actual level of risk until several HGGM attempts have been conducted. For utilitarians, the presence of some risk should not in itself be an obstacle to the pursuit of HGGM. Nevertheless, opponents of GM often present risk—especially unknown risk—as grounds for absolute proscription. In this context, the ‘precautionary principle’ (PP) is frequently deployed, explicitly or implicitly, against proposals to develop or deploy new genetic technologies, including HGGM.85
The PP is intuitively attractive to many, as it appeals to the deep human heuristic favouring caution. However, the PP implies unacceptably negative, or even absurd, consequences. The strongest version of the PP holds any degree of risk to be unacceptable, irrespective of the potential benefits. Unfortunately, it is impossible to do anything innovative in bioscience or medicine without some risk, or that could eventually lead to undesirable outcomes.86 Various attenuated forms of the PP have been proposed, which seek to avoid the absurd implications of the strong PP.87 However, such variants inject nonconsequentialist elements—such as the allocation of capriciously high weightings to certain types of risk, for example ‘severe’ or ‘unknown’ risks—into the assessment of risk versus benefit. Utilitarians are bound to reject such intrusions, as they can only lead to a skewing of the utilitarian calculus and a concomitant bias towards prohibition.
Another PP‐based objection against proceeding with a germline GM attempt focuses not on risks to the immediate children who may be created with the initial attempts at germline GM, but on the spectre of distant harms or other negative outcomes. On this view, a ‘slippery slope’ is perceived that would lead inexorably from an early‐use to a dystopian situation, with the corollary that it would be unethical to ever attempt germline GM, lest it lead to such eventual disaster. While the general concept of a slippery slope is intuitively attractive to many laypeople and media commentators, it is supported by few ethicists.88 This is because the concept lacks a logical basis: why should the first use of any novel biomedical technology inexorably lead to disaster? I suggest that, to qualify as a coherent argument, the onus resides with the proponents of a slippery slope to establish a plausible and strong link between the early use of a new technology and the evoked dystopian future.
Of course, possible future utility‐reducing downsides of widespread HGGM can be envisioned, including negative societal effects through genetic inequity, opportunity costs through diversion of resources from alternative projects, and a narrowing of the human gene pool when the technology becomes able to alter large numbers of sequences per embryo. And it has been argued that, just as a slippery slope is a fallacious assumption, it would be unwarranted to assume the existence of a simplistic ‘automatic escalator’ in which technological improvement leads to some utopia.89 However, history demonstrates that progress in medical science and technology has indubitably produced major net benefits for humanity; and there is every reason to believe that further major gains will accrue, commensurate to the extent that bioscientific research is allowed to proceed.90 The posited downsides lie far in the future; their probability and magnitude is highly uncertain, and technological or legislative solutions may well emerge to prevent or mitigate them. To proscribe a promising biotechnological development in its initial stages based on possible far‐future downsides would be irrational and unconducive to utility maximization.
13 A PRUDENTIAL UTILITARIAN APPROACH
In the farther future, HGGM is likely to become highly effective and efficient: it is conceivable that it might ultimately be applied to large numbers of people, perhaps in a manner akin to present‐day immunization programmes, thereby preventing millions of premature deaths and a great deal of suffering. We cannot know whether the necessary technological developments to reach this stage will take 15, 50 or 100 years, or more: but it is clear that a delay to the start of this journey would postpone the future benefits. If we wait for (say) 10 years before proceeding, that will be 10 years lost, which equates to a substantial forfeiture of utility. Of course, the people who stand to benefit the most from HGGM do not yet exist; but the happiness of these future people must be included in the utilitarian calculus.
Thus, a delay in implementing the technology would be a source of disutility. This does not necessarily mean, however, that it must be right to permit HGGM immediately, without any delay whatsoever. An alternative approach would be to wait a modest period—perhaps 1 or 2 years—to allow the underlying technology to mature further. This delay would entail some utility loss, because some couples would have more difficulty conceiving later, and some would opt for natural conception in the meantime. But it would also bring several advantages. Efficacy and precision will likely further improve as the basic technology is refined, meaning that the risks of both failure (and the associated burden of IVF cycles) and genetic damage (and associated disease) should be reduced.
While these benefits from a delay would boost utility, the magnitude of the utility gain will be low, considering that (as discussed above) current CRISPR technology already offers the ability to reliably alter genetic sequences with relatively low levels of genetic damage. Additionally, a delay that appeared to be based on acceding to the notion that the technology is insufficiently safe might inadvertently lend support to arguments for an indefinite delay, as it will never be possible to know with certainty what the actual safety levels are until the technology is used to create human children. However, a modest delay might offer the pragmatic benefit of allowing public trust to build. If more cellular and in vivo (animal) reports of success are allowed to accumulate, and thus counter residual safety concerns (such as the above‐mentioned negative report of Kosicki et al.), this may serve to provide reassurance to the public, thereby creating greater acceptance for HGGM.
The recent news that He Jiankui has seemingly created gene‐edited human infants is of particular relevance to the question of a delay. As discussed above, this first attempt at HGGM looks ill‐conceived and early indications are that it has not been fully successful; it is thus likely to function as a setback for the start of the HGGM revolution.91 If another attempt, such as that suggested for Couple X, were also to fail badly, it seems likely that public support for the whole enterprise would be severely impacted, leading to long delays in going forward with HGGM. Thus, although it can be argued that a properly conceived GM attempt, such as that described for the Couple X scenario, might justifiably proceed straightaway, the bad publicity surrounding the first actual attempt at HGGM gives further support to the argument that utilitarians should support a modest delay, of perhaps 1 or 2 years, before moving ahead with further attempts to modify the human germline.
A permanent or long‐term prohibition on the use of GM technology would be antithetical to progress in biomedical innovation, and hence unethical. Although there exist residual safety concerns about CRISPR, there are strong grounds to believe that the technology could be used to create a GM child at an acceptable level of risk. It is possible to identify plausible cases where the application of HGGM technology, at its current stage of technical development, would bring about direct beneficial consequences for those concerned. Couple X in the foregoing discussion represents a hypothetical candidate case, involving SCA. A successful CRISPR‐based HGGM attempt would, despite the burden of the process on the mother, boost utility in respect of these parents, and their future child, if the child would otherwise have been conceived naturally and thus destined to develop SCA. However, the availability of HGGM may lead to some parents choosing this approach instead of adopting a child, which would be a disutility as the happiness of orphaned children can be greatly increased through adoption. But a successful early‐use attempt would have the wider benefit of serving to kickstart the whole enterprise of HGGM. This means that even if there was an initial net reduction in utility due to reduced adoption associated with the early use of GM, it would most probably be outweighed by the utility‐generating effects of bringing forward the development of advanced, widely applicable HGGM technology. From a utilitarian perspective, no principled reasons exist to support a risk‐averse ‘precautionary’ delay on an early‐use HGGM attempt. However, a modest delay would have pragmatic benefits, a notion that has been given added impetus by the recent news of an ethically questionable and apparently not entirely successful first attempt at HGGM. I suggest that utility will be maximized if we kickstart the next biomedical revolution by proceeding not immediately but within around 1–2 years to intervene in the human germline.
CONFLICT OF INTEREST
The author declares no conflict of interest.
- Kevin Smith is a bioethicist at Abertay University, Dundee. He has a scientific background in genetics, focusing on genetic modification (GM). In addition to ‘genethics’, his major area of interest is in the ethics of anomalous therapies. He has published a number of academic papers, book chapters and books spanning these scientific and bioethical areas.