Chapter 2 described what is currently achievable through assisted reproductive technologies (ARTs) and the state of the science relevant to genome editing to prevent the transmission of a heritable disease. Chapter 2 concluded that there are significant gaps in scientific knowledge that would need to be filled before heritable human genome editing (HHGE) could responsibly be considered for clinical use. Chapter 3 addresses questions raised by defining a responsible clinical translational pathway for evaluating a potential use of HHGE, in the event that a country chooses to do so. After laying out general considerations related to the potential harms, benefits, and uncertainties of HHGE, Chapter 3 outlines six broad categories of uses for which HHGE could be considered and describes the genetic and clinical considerations associated with each category. It then sets out the Commission’s conclusions about the circumstances in which a translational pathway could responsibly be described for HHGE and explains why it is not possible at present to describe a responsible translational pathway for other types of potential use.
Decisions about the clinical use of HHGE involve issues that are complex and arguably unprecedented, because the potential range of modifications that could be made to the human genome is vast (from correcting a known disease-causing mutation to inserting new genes or regulatory
elements); the effects are potentially multigenerational; and the potential scope of purposes that might someday be considered is wide-ranging, from helping prospective parents who currently have no prospect of having a genetically-related child unaffected by a severe genetic disease to, in the extreme, engaging in a program of eugenic modification of the human species. Moreover, the range of people who might experience potential benefits and harms is extensive, including prospective parents, their offspring, future generations who inherit these genetic modifications, and the society at large.
The clinical use of HHGE requires addressing two distinct issues: (1) whether a country decides that the clinical use of HHGE is appropriate for any purpose and, if so, for which purposes; and (2) if a purpose is deemed appropriate, how to define a responsible translational pathway for evaluating it with respect to efficacy and safety.
The first issue involves societal values, including ethical, cultural, legal, and religious considerations, and should be informed by scientific knowledge. Decisions about whether to permit clinical use of HHGE at all will involve weighing both deep concerns (ranging from the appropriateness of altering human DNA, regarded by some as a fundamental aspect of humanity, to the ultimate societal impact of widespread and extensive manipulation of the human genome) and deep obligations to ensure that humankind can benefit from scientific knowledge and medical advances. Decisions about the appropriateness of specific uses may depend on the extent to which they are judged to address a compelling need. Interest in using HHGE as an ART—to assist at-risk couples to have a genetically-related child that does not inherit a serious genetic disease—is driven by the recognition that many couples have a strong preference to have children who are genetically-related to both parents (Hendriks et al., 2017; Rulli, 2014; Segers et al., 2019). In contrast, interest in using HHGE to “enhance” the human species involves different motivations and raises serious issues associated with discredited projects of eugenics. Consideration of HHGE will thus need to involve careful societal decision making about whether and when to cross thresholds, informed by scientific knowledge but relying on value judgments. While very important, such considerations lie beyond the remit of this Commission.
The second issue is at the heart of the Commission’s task: defining responsible translational pathways for particular uses of HHGE, should a country judge the uses appropriate. Defining responsible translational pathways clearly involves scientific considerations, but it also entails societal and ethical considerations related to weighing potential benefits and harms, and uncertainties about them, in the clinical evaluation of a new medical technology. Notably, the Commission’s Statement of Task requires considering both the research and clinical issues and the “societal and ethical issues, where inextricably linked to research and clinical practice.”
Below, the Commission considers the circumstances for which it concludes it is currently possible to define a responsible translational pathway only for initial clinical uses of HHGE, should a country choose to permit them. Decisions to go beyond these initial uses would depend on scientific conclusions based on experiences gained from the initial uses, societal decisions about the appropriateness, and the definition of responsible translational pathways.
CRITERIA FOR DEFINING RESPONSIBLE TRANSLATIONAL PATHWAYS FOR INITIAL USES OF HERITABLE HUMAN GENOME EDITING
A number of considerations are common to the initial human use of new biomedical technologies: the prioritization of safety, very careful selection of a small number of initial cases, emphasis on a favorable balance of risks and potential benefits, and careful review of initial results prior to additional uses. New interventions, with necessarily high degrees of uncertainty about their efficacy, tend to focus on diseases and individuals for whom there are no available alternatives, and on diseases or conditions for which mortality is high and/or morbidity is severe, thereby reflecting the most favorable balance of potential harms and benefits. When these considerations are met, participation in first-in-human uses can be offered to candidates after a process of informed consent.
These three considerations apply to initial uses of HHGE, along with two additional ethical issues that further support attention to the three considerations. First, because HHGE would be a reproductive technology, and as is the case in the use of any ART, prospective parents can provide consent, but the individual who would be created as a result of the technology cannot provide consent. Second, HHGE would create a heritable genetic alteration, which could be passed to future generations. This collection of considerations supports the criteria outlined below.
The Commission’s approach was also informed by analyses undertaken by groups that evaluated the acceptability and considerations for undertaking the initial uses of mitochondrial replacement techniques (MRT) in humans (e.g., Bioethics Advisory Committee, Singapore, 2018; HFEA, 2016; NASEM, 2016; NCB, 2012). Although a technology with more limited potential scope of application than HHGE, MRT provides a useful starting point, given its parallels as a novel ART that creates heritable genetic changes with the aim of enabling parents to have a genetically-related child unaffected by a disease. A U.S. National Academies of Sciences, Engineering, and Medicine study concluded that “[i]n assessing the ethics of the balance of benefits and risks in MRT clinical investigations, minimizing the risk of harm to the child born as a result of MRT is the primary value to
be considered,” reaching this conclusion through “an approach that entails weighing, first and foremost, the probability of significant adverse outcomes borne by the children born as a result of MRT against the benefits accruing to families desiring children who are related to them through their [nuclear] DNA” (NASEM, 2016, p. 115–117). In addition, the United Kingdom gave permission to consider MRT for initial human uses only for circumstances in which preimplantation genetic testing (PGT) was very unlikely to enable prospective parents to have a genetically-related child without a serious mitochondrial disease, and under very strict regulatory oversight (see Chapter 1 for further detail on the regulatory oversight and licensing system for MRT in the United Kingdom).
The Commission distilled these general considerations into a set of principles to guide the development of a translational pathway for any initial use of HHGE:
Highest priority on safety. A combination of factors all support the need to assure the highest level of safety in initial uses of HHGE. As an ART, HHGE would be directed to creating a person without a specific genetic disease (as is the case with PGT), rather than treating an existing patient with a disease.1 The level of safety considered acceptable for permitting initial clinical uses of HHGE should consequently be considerably higher than for somatic cell genome editing.
Most favorable balance of potential harms and benefits. In order to create the most favorable balance of potential harms and benefits, new interventions with substantial uncertainties ideally focus on diseases and prospective parents for whom there are no available alternatives, and on diseases or conditions for which mortality is high and/or morbidity is severe. The use of seriousness of a condition as a criterion for medical intervention is common in laws, regulations, and policy statements (Kleiderman et al., 2019; Wertz and Knoppers, 2002) and the seriousness of the disease in question is currently a central consideration for other ARTs that aim to prevent transmission of a heritable disease, such as PGT and MRT. Although there is not a uniformly established definition of what clinical
1 In a clinical context, the nature and extent of harm is most commonly understood by comparing the health of an individual before and after a treatment. Understanding harms (and benefits) in the context of an ART such as HHGE involves comparison between the health of an individual born following genome editing of a given embryo with the anticipated health of the individual that would have been born if the same embryo had been transferred but without the prior use of genome editing. While recognizing that this raises some philosophical questions, the Commission felt that the word ‘harm’ intuitively captures the idea that individuals born following HHGE could be unintentionally and negatively impacted at some point by the editing procedure.
presentation is meant by “serious,” the concept tends to reflect general ideas that the effects are severe or life threatening or entail substantial impairment. For the purposes of HHGE, the Commission provides below a definition of “serious disease.”
Minimizing potential harms resulting from the intended edit. It is important that the consequences of the intended genome edit are well understood, both for the immediate offspring and for future generations who might inherit it, in order to be certain that an intended edit would not have unintended deleterious consequences (on its own, via genetic interactions with other loci, or via environmental interactions). At present, the best way to achieve this goal is for an edit to change a known pathogenic genetic variant responsible for a monogenic disease to a sequence that is common in the relevant population and known not to be disease-causing.
Minimizing potential harms resulting from unintended edits. To minimize the potential for harm, it is important to minimize the chance of unintended on-target and off-target edits, which could be passed to future generations, as well as indirect effects of the editing process that could affect embryo viability or developmental potential.
Minimizing potential harms by preventing genome editing when there is no prospect of benefit. For the vast majority of prospective parents at risk of transmitting a genetic disease, only a fraction of their offspring would inherit the disease (typically, 25–50 percent). It is crucial to ensure that individuals are not created by genome editing of zygotes or embryos that do not carry the disease-causing genotype, because such individuals would have been exposed to the risks of HHGE without offsetting benefits offered by the procedure.
Availability of alternatives to HHGE that might enable parents to have a genetically-related child unaffected by a specific disease. A key consideration is whether the prospective parents already have reasonable options for conceiving a genetically-related child who does not inherit a serious genetic disease. To maximize potential benefit and minimize potential harm, it would be appropriate to confine initial uses to prospective parents who lack viable options.
Based on the above principles, the Commission identified four criteria that should be met by any proposed initial uses of HHGE, in the event
that a country chooses to permit them. These criteria together emphasize safety with respect to the resulting individual and an acceptable balance of potential harms and potential benefits to that individual:
- The use of HHGE is limited to serious monogenic diseases; the Commission defines a serious monogenic disease as one that causes severe morbidity or premature death.
- The use of HHGE is limited to changing a pathogenic genetic variant known to be responsible for the serious monogenic disease to a sequence that is common in the relevant population and that is known not to be disease-causing.
- No embryos without the disease-causing genotype will be subjected to the process of genome editing and transfer, to ensure that no individuals resulting from edited embryos were exposed to risks of HHGE without any potential benefit.
- The use of HHGE is limited to situations in which prospective parents (i) have no option for having a genetically-related child that does not have the serious monogenic disease, because none of their embryos would be genetically unaffected in the absence of genome editing; or (ii) have extremely poor options, because the expected proportion of unaffected embryos would be unusually low, which the Commission defines as 25 percent or less, and have attempted at least one cycle of PGT without success.
The Commission concluded that a responsible translational pathway for initial uses of HHGE would need to meet all four of these criteria.
For the application of these four criteria in any specific proposed clinical use of HHGE, a case by case evaluation of potential risks and benefits would be required and should proceed under appropriate regulatory oversight.
To apply these criteria in practice, this section outlines possible categories of use of HHGE. The Commission identified six broad categories of potential uses of HHGE, which depend on the nature of the disease, its pattern of inheritance, and other criteria.2 The specific diseases cited under each category are only intended as examples.
2 It has been suggested that HHGE could be used to increase the chances of having a “savior sibling” (a child that is a suitable immunological match for an existing child requiring an organ or cell transplant, as discussed in Chapter 2). The Commission does not discuss this
Category A: Cases of Serious Monogenic Diseases in Which All Children Would Inherit the Disease Genotype
Disease: Serious monogenic disease, with high penetrance.
Genome editing: Change a well-characterized pathogenic variant to a common, non-disease-causing sequence present in the relevant population.
Circumstances: Couples for whom all children would inherit the disease-causing genotype. These circumstances include:
- Autosomal dominant disease. If one parent carries two disease-causing alleles (affected homozygote),3 all children would inherit the disease-causing genotype.
- Autosomal recessive disease. If both parents carry two disease-causing alleles in the same gene (affected homozygotes), all children would inherit the disease-causing genotype.
- X-linked recessive diseases. If the prospective female parent carries two disease-causing alleles (affected homozygote) and the male parent carries a disease-causing allele on his only X-chromosome (affected hemizygote), all offspring would be affected.
The circumstances in this category are rare for two reasons.
From a probabilistic standpoint, the circumstances involve couples carrying more disease-causing alleles than typical. The circumstances can arise where there is an unusually high frequency of a disease-causing mutation in a population, a high prevalence in a population of couples who are close relatives (consanguinity), or a tendency of individuals with the disease to meet and reproduce (assortative mating). The prevalence of circumstances in Category A is discussed in more detail later in this chapter.
From a medical standpoint, the circumstances in this category apply only to a small minority of instances. Because the circumstances involve one or both parents being affected with disease, they arise only for those
possibility further since it does not satisfy two of the Commission’s four criteria for initial uses: (1) Genome editing of the histocompatibility locus would not be an example of changing a known pathogenic gene sequence (and would also be technically very challenging, involving the need to edit multiple genes); and (2) the savior sibling would be effectively exposed to the risks of HHGE, but the benefit would accrue to someone else.
3 Individuals carrying two disease-causing alleles are referred to as homozygous if the two disease-causing mutations are identical and as compound heterozygous if the two disease-causing mutations are different. Aside from the discussion in Chapter 2 of the added complexity of editing multiple alleles in compound heterozygotes, this distinction does not matter for the purposes of this chapter, and the term homozygous will be used throughout.
serious monogenic diseases that are compatible with individuals surviving to reproductive age with preserved fertility.
Examples of serious monogenic diseases for which the circumstances in this category can arise include autosomal dominant diseases such as Huntington’s disease; and autosomal recessive diseases such as cystic fibrosis (CF), sickle cell anemia, and beta-thalassemia.
Considerations: This category is unique in two important respects. First, for couples in this category prenatal diagnosis and PGT, which can identify fetuses and embryos that have not inherited the disease-causing genotype, have no chance of identifying genetically unaffected embryos. Second, the embryos exposed to risks associated with genome editing procedures would be only those carrying the disease-causing genotype for a serious monogenic disease. This stands in contrast to the situation in Category B, below.
Category B: Serious Monogenic Diseases in Which Some, but Not All, of a Couple’s Children Would Inherit the Disease-Causing Genotype
Disease: Serious monogenic disease, with high penetrance.
Genome editing: Change a well-characterized pathogenic variant to a common, non-disease-causing DNA sequence present in the relevant population.
Circumstances: Couples for whom some children would inherit the disease-causing genotype. The typical circumstances are:
- Autosomal dominant disease. If one parent carries one copy of a disease-causing allele (affected heterozygote), on average 50 percent of children would inherit the disease-causing genotype and 50 percent would not.
- Autosomal recessive disease. If both parents are unaffected heterozygous carriers for disease-causing alleles, on average 25 percent of children would inherit the disease-causing genotype and 75 percent would not.
- X-linked dominant disease. If the mother is heterozygous for the disease-causing allele, on average 50 percent of all children would inherit the disease-causing genotype.
- X-linked recessive disease. If the mother is a heterozygous carrier and the father does not carry the disease-causing allele, 50 percent of male offspring on average would inherit the disease-causing genotype. Fifty percent of female offspring will be heterozygous carriers and will typically be clinically unaffected, although exceptions some
times occur due to skewed inactivation of the genetically unaffected X chromosome, resulting in females with varying manifestations of disease.
In rare circumstances, the expected proportion of affected offspring can be higher. If both parents are heterozygous for a disease-causing allele for an autosomal dominant disease, on average 75 percent of children would inherit the disease-causing genotype. If one parent is an affected homozygote for an autosomal recessive disease and the other is a heterozygous carrier, 50 percent of the children would be expected to inherit the disease-causing genotype. These instances are expected to occur more frequently in populations with “founder effects” in which the contemporary population is derived from a small founding population, resulting in reduced allelic diversity, with particular disease-causing alleles persisting at relatively high frequency in the population.
Couples in Category B are far more common than couples in Category A, for two reasons. From a probabilistic standpoint, Category B typically involves individuals carrying one rather than two disease-causing alleles for a given disease. From a medical standpoint, Category B comprises many more diseases than Category A. Because the parents in Category B may be unaffected carriers in the case of autosomal and X-linked recessive diseases, the category includes all of the thousands of serious recessive and X-linked diseases. In contrast, because Category A involves only affected parents, it includes only the small subset of serious diseases for which affected individuals can survive to reproductive age.
The proportion of all reproductive couples that fall in Category B is substantial. The World Health Organization estimates that a monogenic disease is present in 1 percent of global births (WHO, 2019b). Only some of these instances fall into Category B, because some monogenic diseases do not meet the Commission’s definition of being serious for the purpose of defining a responsible translational pathway for HHGE, and some couples have affected children due not to inherited mutations but to newly arising (de novo) mutations, which by definition could not have been prospectively identified in the parents. The Commission estimates that Category B comprises at least 0.1 percent of all couples, estimated at more than 1 million couples worldwide.
While most cancers arise from a constellation of somatic mutations, there are individuals with inherited cancers due to single mutations with high penetrance, which in some cases can be prevented by surgical removal of target tissues or organs. For example, familial adenomatous polyposis has virtually 100 percent penetrance without surgical removal of the colon. In the setting of very high penetrance, such cancer syndromes could be considered serious inherited diseases. Other inherited cancer syndromes
feature lower penetrance, and other considerations would need to be taken into account in an assessment of potential harms and benefits. Additionally, other inherited variants may make more modest contributions to cancer risk, as discussed below in Category C or D.
Considerations: Category B differs from Category A in two important respects.
First, the application of HHGE, as currently conceived, would involve treating all zygotes at the single-cell stage, regardless of their genotype. Category B would therefore involve subjecting all embryos to risks associated with genome-editing procedures—including those that do not have the disease genotype and thus do not require genome editing. Possible alternative approaches for the application of HHGE that would avoid editing unaffected embryos are discussed below.
Second, couples in Category B currently have existing options for having an unaffected child genetically-related to both parents—specifically, embryo selection via PGT. The limitation here is quantitative rather than qualitative. For any given couple, the probability of success in producing a child is somewhat lower for PGT than for in vitro fertilization (IVF) in general. As noted in Chapter 2, it is estimated that approximately 80 percent to 90 percent of PGT cycles in which at least one embryo reaches the diagnosis stage result in an unaffected embryo that can be transferred to the uterus. The remaining 10 percent to 20 percent of PGT cycles do not result in transfer of an unaffected embryo, and some couples may not obtain an unaffected embryo even after several PGT cycles (particularly those from whom only a small number of eggs can be harvested).
It has been proposed that, if HHGE worked with high efficiency and safety, it might assist certain couples (those who currently have a small number of unaffected embryos) by increasing the proportion of unaffected embryos available for uterine transfer. However, this outcome is by no means certain, because the additional laboratory procedures involved in HHGE might decrease the yield of high-quality embryos available for transfer.
Category C: Other Monogenic Conditions with Less Serious Impacts Than Those in Categories A and B
Disease: Monogenic disease or disability with less serious impacts than those in Categories A and B.
Genome editing: Change a well-characterized pathogenic variant to a common, non-disease- or non-disability-causing sequence present in the relevant population.
Circumstances: This category involves prospective parents for whom all or some of their naturally conceived children would inherit the genotype that causes the monogenic condition.
An example is familial hypercholesterolemia (FH), which is caused by mutations in the gene encoding the low-density lipoprotein (LDL) receptor (or mutations in a number of other genes such as APOB and PCSK9) and can occur in heterozygous or homozygous form. Heterozygous FH is a relatively common genetic condition (with a frequency of approximately 1 in 250) that causes elevated LDL cholesterol levels that predispose to early cardiovascular morbidity and mortality. Moreover, LDL levels in these individuals can usually be effectively reduced with medications, substantially lowering the risk of heart attack, reduced quality of life, and premature death. By contrast, homozygous FH, which is rare (with a frequency of around 1 in 160,000–300,000 globally, but higher in populations with FH founder variants) causes an extreme form of hypercholesterolemia that is difficult to treat and typically leads to life-shortening heart disease (Cuchel et al., 2014), although new therapeutic approaches are being developed to effectively treat these patients. The use of HHGE in a case in which both parents carry a disease-causing FH allele would fit in Category B. When only one parent is a carrier, the case would belong in Category C, since only heterozygous or unaffected embryos could result.
A second group of examples involves genotypes that may affect an individual’s quality of life but are not serious monogenic diseases within the meaning of the Commission’s definition (a disease that causes severe morbidity or premature death). Inherited deafness would be an example. While some deaf individuals consider deafness as severely impacting quality of life and a condition to be avoided, others strongly disagree (Padden and Humphries, 2020). The Commission recognizes that a country’s consideration of genome editing for conditions such as deafness raises many complex issues that are beyond this report’s scope.
Considerations: Although Categories B and C both comprise monogenic disorders, compared with Category B the conditions in Category C feature less severe morbidity, and risk of premature death may be mitigated by relatively simple medical or lifestyle interventions.
Category D: Polygenic Diseases
Disease: Polygenic diseases, for which a large number of genetic variants each contributes to disease risk, with the variants collectively having substantial—though not determinative—effect on disease occurrence or severity.
Genome editing: Changing one, several, or even large numbers of genetic variants associated with higher risk of the disease to alternative common variants that are associated with lower risk of the disease.
Circumstances: The risk of developing common diseases is influenced by many genetic variants (often hundreds or more), as well as by interactions with non-genetic factors collectively referred to as environment (which may include diet, pathogen exposure, exercise, and much more). Most of these genetic variants are common alleles that have small effects on disease risk (altering risk by less than a factor of 1.1-fold), although a few common variants can have relatively large effects on this risk and rare alleles in some genes can have large effects on risk of common disease. The combined effects of these risk variants are often additive, although there may sometimes be genetic interactions (i.e., the presence of one variant may alter the effect of another variant).
Examples in Category D include many common diseases, such as type 2 diabetes mellitus, heart disease, and schizophrenia, although rare Mendelian forms of common diseases can occur. For common polygenic diseases, changing a single genetic variant would typically be expected to have negligible effect on the risk of disease. A notable exception is the E4 allele of the APOE gene: the risk of developing Alzheimer’s disease rises with every decade of life after age 60, but risk increases more rapidly depending on whether an individual has zero, one, or two copies of the E4 allele (Rasmussen et al., 2018). Even so, the APOE gene explains only a fraction of the risk for Alzheimer’s disease (absolute risk of ~5 percent from ages 60 to 69, and 15–20 percent over age 80).
Considerations: Current scientific understanding suggests that genome editing to alter one or more gene variants associated with a polygenic disease would be unlikely to prevent the condition and might have undesired effects, as the targeted alleles may play important roles in other important biological functions and may interact with the environment. Moreover, potentially better options may be available or become available to minimize the risks of developing the disease or to help manage its consequences.
Category E: Other Applications
Disease: This category does not involve heritable diseases. Rather, it involves genetic changes directed toward other objectives, which may or may not be health-related and may involve introducing genetic sequences that do not naturally, or only very rarely, occur in the human population.
Genome editing: Genetic changes ranging from single-base substitutions to introduction of new genes or disabling of existing genes.
Circumstances: A vast range of applications can be imagined for HHGE, ranging from attempts to prevent or protect against infectious diseases, to genetic changes that would enhance normal human traits, to introducing genes conferring new biological functions. All of these applications raise scientific, societal, and ethical questions that are impossible to resolve given the current state of scientific understanding. Examples include:
- attempting to provide offspring with resistance to an infectious disease by editing a gene, for example, the attempt to inactivate the CCR5 gene and confer resistance to HIV infection;
- attempting to produce an ability in offspring by introducing a rare allele of a specific gene known or believed to be associated with a desired phenotype. For example, constitutive activation of the EPO gene has been proposed to confer advantages in endurance sports (Brzeziańska et al., 2014);
- attempting to modify traits such as height or cognitive ability that are influenced by hundreds or thousands of genetic variants across the genome; and
- attempting to confer new abilities, not found in humans, by adding sets of genes that, for instance, might confer resistance to radiation exposures encountered during extended spaceflight.
Considerations: In all of the cases mentioned above, the potential impacts of HHGE on children, adults, and future generations cannot be fully assessed. For example, while it is clear that homozygous loss of CCR5 function confers partial protection from HIV infection, this loss may increase other risks of morbidity. Moreover, effective methods of preventing and treating HIV infection are available. Similarly, a lifelong increase of red blood cell mass due to constitutive expression of erythropoietin may increase endurance but might also increase the lifetime risk of thrombosis. For these reasons, the benefit-to-harm ratio in these scenarios is uncertain and in many instances may be very low.
In addition to these scientific and clinical complexities there are, of course, numerous ethical and social obstacles to interventions in this category. Any future justification for pursuing such interventions would require both scientific agreement that the long-term impact of such changes can be assessed and societal approval about the acceptability of such interventions.
Category F: Monogenic Conditions That Cause Infertility
A special category for which genome editing might be used is in treating germline cells (or their precursors) from an existing individual to reverse infertility with a monogenic cause. In this case, genome editing would change the sequence of a gene to restore fertility. Whereas HHGE in Categories A through E would not be directed at offering therapy for an existing individual suffering from a disease but rather would be a form of assisted reproduction, Category F has the unique feature that the intended beneficiary of the genetic alteration would be an existing individual (the infertile prospective parent) with the additional impact that the edited genome would be transmitted to offspring.
This category remains hypothetical for now because, leaving aside the issues of genome editing, it is not currently possible to generate functional gametes from human stem cells. Any developments in this area would require regulatory approval for a range of ARTs before clinical applications could be considered.
The Commission then considered the circumstances within the categories above that could meet the four criteria outlined earlier in the chapter for which a responsible translational pathway could currently be described. Based on this analysis, the Commission concluded that initial uses of HHGE would need to be restricted to Category A and a very small subset of Category B, provided certain conditions can be met. This section discusses Categories A–F in turn.
Category A clearly meets the four criteria for initial uses of HHGE: (1) The category involves serious monogenic diseases. (2) Genome editing would be directed at changing a pathogenic variant known to be responsible for the serious monogenic disease to a sequence commonly carried in the relevant population. (3) No individuals resulting from edited embryos could have been exposed to potential harms from HHGE without potential benefit, because all of the couple’s embryos carry the disease-causing genotype. (4) Couples currently have no other options to produce a genetically-related child free of the disease.
Category B would not, as a whole, be suitable for initial uses of HHGE, because it does not currently meet the third criterion and because most couples would not meet the fourth criterion. The key difference from Category A is that couples in Category B can produce children who do not inherit the disease-causing genotype (in typical cases, at least half on average). With respect to the third criterion, HHGE, as currently conceived, would involve subjecting all zygotes (both those that do and do not have the disease-causing genotype) to genome editing procedures and thus would result in the birth of children derived from embryos that had been needlessly exposed to potential harms from genome editing. With respect to the fourth criterion, the vast majority of couples already have a viable option (PGT) for producing a genetically-related child that is free from the genetic disease. As discussed above, the substantial majority (80–90 percent) of PGT cycles in which at least one embryo reaches the diagnosis stage results in an unaffected embryo that can be transferred to the uterus. The primary interest in HHGE in Category B is to assist couples who have very low prospects of having an unaffected child, owing to few unaffected embryos being available for transfer.
After extensive discussion, the Commission concluded that initial uses of HHGE might be appropriate under certain circumstances for a very small subset of Category B.
First, reliable methods would need to be developed that ensure that no individual would be produced from embryos that had been needlessly subjected to HHGE, ideally by identifying embryos that carry the disease-causing genotype before performing HHGE. One approach might be to use polar-body genotyping, which has the potential to identify zygotes that have inherited from the mother an allele that causes a dominant monogenic disease (see Chapter 2); the reliability of polar-body genotyping for this purpose would need to be established. Those zygotes could be subjected to HHGE followed by PGT, while the other zygotes could be subjected to standard PGT. Another approach might be to develop reliable procedures to perform HHGE on multicellular embryos without producing embryos that are mosaic for the edit. This approach would enable the genotype of an embryo to be determined prior to delivering editing reagents. However, no such procedures are currently available.4
4 In theory, a third approach would be to perform HHGE on all zygotes, subsequently identify by PGT those embryos that did not have the disease-causing genotype prior to genome editing and had therefore been needlessly subjected to HHGE (because HHGE targets the disease-causing mutation(s), this would require genotyping a sufficient set of polymorphic sites on each side of the mutation to distinguish the two haplotypes in each parent), and ensuring that those embryos are not transferred. However, many Commission members viewed this approach as problematic because it would require a commitment to discarding embryos that
Second, initial uses would need to be restricted to those couples that have very poor prospects for having an unaffected child with conventional PGT. The Commission defines such couples as those (i) for whom the expected proportion of unaffected offspring is 25 percent or less (for example, couples in which both parents are heterozygous for the same or different dominant serious monogenic diseases) and (ii) who have undergone at least one cycle of PGT without success, since many couples will produce enough embryos to yield unaffected embryos suitable for transfer without editing.
To meet all four criteria, any initial uses of HHGE in Category B should be confined to these circumstances.
Categories C Through F
Category C involves genetic diseases that have less serious effects, may be manageable using other methods, and may not be seen as negatively impacting quality of life by members of communities affected by the condition. Until much more is known about the safety and efficacy of HHGE, it is unclear that the potential benefits outweigh the potential harms. A cautious approach argues against undertaking first-in-human uses in this category.
Category D (polygenic diseases) and Category E (genetic changes that are not directed toward variants involved in heritable diseases and may involve genetic sequences that do not naturally occur in human populations and uses that could be seen as enhancements) are not currently suitable for HHGE. Scientific understanding and existing technologies are insufficient to produce predictable, well-characterized results, including across a range of genetic and environmental interactions, and to minimize the effects of unknown and speculative risk. Moreover, these uses raise additional societal and ethical concerns.
Category F (monogenic conditions that cause infertility) remains speculative at present, making it impossible to define a responsible translational pathway. Since human stem cell–derived in vitro gametogenesis has not been developed or permitted anywhere for medical use, it is premature to consider how it might be used in combination with HHGE.
would have been suitable for transfer but for the fact that they had been needlessly subjected to potential harms from HHGE.
Circumstances for a Responsible Translational Pathway for Initial Uses of Heritable Human Genome Editing
In summary, the Commission concluded that a responsible translational path for initial uses of HHGE
- could be defined for Category A;
- might be defined for the very small subset of couples in Category B who have a very low likelihood of success through PGT due to genetic circumstances (embryos having a 25 percent or lower probability of not inheriting the disease-causing genotype) and who have attempted at least one PGT cycle without success, provided that reliable methods are established to ensure that no individuals result from embryos that were needlessly subjected to HHGE; and
- cannot currently be defined for the rest of Category B or for Categories C through F.
As previously discussed, prior to any clinical use of HHGE in any circumstances, it will be necessary to demonstrate a safe and effective methodology and for any country offering HHGE to have an appropriate regulatory framework to oversee it. Before crossing the threshold of undertaking clinical uses of HHGE in other circumstances beyond those described above, an appropriately constituted international body should assess whether and under what circumstances a responsible translational path can be defined.
The Commission next considered the frequency of the circumstances for initial uses of HHGE defined above, to determine whether there is likely to be an adequate number of suitable couples to enable initial studies to evaluate efficacy and safety, which we judge to be approximately 10–20 couples. Our analysis suggests that there is likely to be an adequate number of prospective parents to reach this goal.5
As discussed below, prospective parents who might be offered HHGE would likely come from multiple countries. This observation reinforces
5 These initial studies would be evaluated for the safety and efficacy of the editing and the likelihood of a successful pregnancy. This would give crucial information for further studies and it would be essential that information about these outcomes be shared. If these studies did not raise concerns about the safety or efficacy of the HHGE technique then much larger studies would be required to evaluate long-term outcomes for the individuals whose genomes had been edited.
the value of global coordination of any clinical use of HHGE. It would be important to use a clear mechanism, such as an international consortium, to identify potential participants, undertake the genome editing intervention according to the translational pathway described in this report, and evaluate clinical outcomes. Precedents exist for such international coordination and collaboration, such as the International Rare Diseases Research Consortium (Lochmüller et al., 2017), including global coordination of clinical trials.
How Common Are the Circumstances in Category A Expected to Be?
The circumstances in Category A are very rare. This is appropriate for the initial use of a technology such as HHGE, where it would be suitably cautious to begin with a small number of couples who have no alternatives, proceed carefully, and intensively study the results. It is important to assess whether there is a sufficient number of couples in Category A that could potentially benefit from HHGE. As noted above, Category A arises only for the minority of serious monogenic diseases that are compatible with individuals surviving to reproductive age and being able to reproduce. Examples of diseases where this is the case are Huntington’s disease, CF, sickle cell anemia, and beta-thalassemia.
The actual number of couples in Category A is not known, although there are anecdotal examples. Basic principles of population genetics can provide an initial insight into the expected frequency of couples in Category A. Under the classic assumption of a closed, randomly mating population (specifically, individuals choose partners from within the population, their choice is not correlated with relatedness or disease status, and disease status does not affect fertility), the expected proportion of couples in Category A will be approximately 2q2 for an autosomal dominant disease and q4 for an autosomal recessive disease, where q is the frequency of disease-causing alleles.6
The frequency of disease-causing alleles differs among diseases, depending on the rate of appearance of new mutations that give rise to new disease-causing alleles and the rate of their removal from the population via natural selection. Alleles that cause serious dominant diseases are typically much rarer than alleles that cause recessive diseases because the latter are only subjected to negative selection when an individual contains disease-causing alleles on both chromosome copies, while dominant alleles are virtually always under negative selection because only one mutant copy is
6 The frequency of homozygotes is q2. For an autosomal recessive disease, both parents must be homozygous (q2 × q2). For an autosomal dominant disease, either parent in a Category A couple may be homozygous (approximately q2 + q2).
needed to produce disease. The collective frequency of all disease-causing alleles in a gene (q) is often in the range of 4.5 × 10–3 for a serious autosomal recessive disease and 2 × 10–5 for a serious autosomal dominant disease.7 From these values, the expected frequency of couples in Category A occurring by chance for a particular gene would be expected to be in the range of 4 × 10–10 for a recessive disease and 8 × 10–10 for a dominant disease — that is, in the range of 4–8 per 10 billion for any given disease gene. If there were 100 similar genes in this category, the total frequency of couples in Category A would be about 100-fold higher (about 4–8 per 100 million couples). Applying similar reasoning, a recent article estimated that there would be only a small number of births from Category A circumstances in the U.S. population (Viotti et al., 2019).
The actual frequency of couples in Category A is expected to be significantly higher in populations that have much higher frequencies of certain disease alleles. For recessive monogenic diseases, an allele frequency of around 3 to 10 percent would correspond to couples in Category A occurring at frequencies between 1 in 10,000 and 1 in 1.2 million. In populations with high rates of consanguineous unions (couples who are closely related genetically) and with local variation in allele frequency, the frequency of homozygotes will be higher and therefore so will the expected frequency of couples in Category A. For dominant monogenic diseases, an allele frequency of 0.1 to 1 percent would correspond to a homozygote frequency between 1 in 10,000 to 1 in 1 million.
On the other hand, when estimating the frequency of couples in Category A, one must take into account the fact that some serious monogenic diseases shorten lifespan or decrease fertility, and some autosomal dominant diseases have more severe disease manifestations in homozygotes than heterozygotes (Homfray and Farndon, 2015; Zlotogora, 1997).
Beyond estimating disease-allele frequencies, another consideration is that, for certain recessive diseases, heterozygous carriers enjoy a benefit in certain environments. This is the case for sickle cell disease (SCD) in areas where malaria is prevalent. In such areas, people with one sickle cell allele who contract malaria are less likely to die from the disease (Archer et al., 2018).
For any disease, it is important to consider whether it is technically feasible to reliably edit the disease-causing mutation. Huntington’s disease, for example, is caused by an expanded number of trinucleotide repeats within
7 Under mutation-selection balance in a randomly mating population, the equilibrium frequency q is expected to be (µ/s)1/2 for a recessive disease and µ/s for a dominant disease, where µ is the mutation rate of new disease-causing alleles and s is the selection coefficient against the affected genotype. The values of µ and s depend on the gene. The figures cited in the text correspond to a mutation rate of new loss-of-function alleles of µ= 10–5 for a ‘typical’ human gene and a selection coefficient s = 1/2.
the gene. HHGE would require reducing the number of these repeats to a non-disease-causing level, which is technically more difficult than changing a single nucleotide. Alternatively, genome editing could be used to introduce sequences that do not naturally occur in the human population (e.g., a stop codon to inactivate the gene); however, our second criterion (above) restricts initial uses of HHGE to producing naturally occurring alleles that are common in the relevant population.
The examples of Huntington’s and sickle cell anemia demonstrate that even among serious monogenic diseases where affected individuals survive to an age when they could have children, there are genetic and environmental factors that complicate the analysis of potential harms and benefits arising from HHGE.
Potential Examples of the Circumstances in Category A
Actual data are not readily available from the literature on numbers of couples of reproductive age in Category A. Nevertheless, as noted in the section above, very approximate estimates may be generated under simplified assumptions of random mating. The estimates suggest that couples in Category A for recessive monogenic diseases may occur at meaningful frequencies in populations with disease-allele frequencies exceeding about 3 percent, with the frequency being even higher in populations with higher rates of consanguinity. Moreover, the number of couples in Category A for dominant monogenic diseases will depend on the frequency of individuals who are homozygous for the disease-causing alleles and can and wish to have children. The following are some examples where there may be a substantial number of couples in Category A.
Beta-Thalassemia in Global Populations
Beta-thalassemia is an autosomal recessive blood disorder that disrupts the formation of hemoglobin and can cause severe anemia and other issues. Patients who produce no functional beta globin (beta-thalassemia major) require regular blood transfusions; those who produce beta globin with significantly reduced function can exhibit a range of disease severity. Without access to regular treatment, thalassemia major patients may die in adolescence, but with improved medical care, life expectancy has risen into the 40s and 50s. Mutations that cause thalassemias are relatively common, with approximately 1.5 percent of the global population estimated to be heterozygous carriers for beta-thalassemia (i.e., up to 80 million people), with high carrier rates noted across the Mediterranean region, Middle East, India, Southeast Asia, and Pacific Islands (De Sanctis et al., 2017). For example, it has been estimated that 4.5 percent of the Malaysia population are carriers
for beta-thalassemia (George, 2001), indicating an allele frequency of 2.25 percent. In a population of roughly 32 million this suggests approximately 10 homozygous couples in that country. In India, the carrier rate is estimated to be between 3 to 4 percent (GUaRDIAN Consortium et al., 2019), indicating allele frequencies of 1.5 to 2 percent. In a population of roughly 1.35 billion people, this suggests there may be between approximately 70 and 200 homozygous couples for beta-thalassemia. In North African countries, estimates of beta-thalassemia carrier rates range from 1 to 9 percent, suggesting allele frequencies in the range of 0.5 to 4.5 percent (Romdhane et al., 2019). Considering just the North African population of about 240 million, the expected frequency of homozygotes is sufficiently high (approximately 1 in 500 to 1 in 40,000, depending on the region) that there may be many couples with both members homozygous for beta-thalassemia.
Sickle Cell Disease in Sub-Saharan Africa and the United States
SCD is an autosomal recessive disorder occurring when an affected individual carries two copies of the allele for sickle cell trait. The prevalence of sickle cell trait is high in many populations in sub-Saharan Africa, due to the heterozygote advantage described above. In one example, screening of several thousand women of child-bearing age and their male partners in the Enugu state of Nigeria (population 3.3 million)8 identified sickle cell trait in 22 percent of individuals (Burnham-Marusich et al., 2016). Based on this frequency, the authors expected to identify approximately 1 percent of their study cohort as having SCD but identified only 0.1 percent of their study cohort as being SCD homozygotes; they speculated that this may be due to early mortality, which has been estimated to be 50 to 90 percent for SCD in sub-Saharan Africa. This rate of reproductive-age SCD homozygotes would suggest a frequency of couples in Category A of approximately 1 per 1 million in this population. The situation is expected to be similar in the many other areas of sub-Saharan Africa in which sickle cell trait is common, suggesting that there could be hundreds to potentially thousands of homozygous couples across those areas where SCD is most prevalent. The frequency of sickle cell trait in the African American population is also relatively high (estimated at roughly 7 percent),9 with more than 90 percent of SCD homozygotes estimated to live past age 18 and commonly into their 40s (Platt et al., 1994; Quinn et al., 2010). Viotti et al. (2019) used this carrier frequency to estimate that there are approximately 80 homozygous couples among African Americans.
The carrier frequency for a mutation in the gene that causes CF, an autosomal recessive disease, is approximately 1 in 30 (around 3 percent) in Caucasian Americans (Strom et al., 2011), resulting in CF in approximately 1 in 3,600 births. Similar estimates for incidence of CF are reported for European populations (Farrell, 2008). The authors of a recent paper estimated that there are only 1–2 reproductive-age couples in the United States in which both parents are homozygous for CF (Viotti et al., 2019). Based on similar CF allele frequencies and the roughly 1.5 times greater population, one could expect several such couples in Europe. The rapid advances in the treatment of CF may result in an increased number of couples in which both people are affected by CF being able to have children.
How Common Are the Circumstances in the Subset of Category B Expected to Be?
To fit the circumstances of the very small subset of Category B, both prospective parents would need to be heterozygous for the same or different serious dominant disease(s). Such circumstances are expected to be rare as they depend on both parents carrying disease-causing alleles and on people with the disease surviving to reproductive age and being able to have children. Some examples of diseases that might be compatible with these circumstances are Huntington’s disease, early onset Alzheimer’s disease, and familial adenomatous polyposis.
Huntington’s is a neurodegenerative disease that arises from an expanded number of three nucleotide repeats in the DNA sequence of the gene HTT. The disease is found in approximately 3–7 per 100,000 people of European descent10 and has been estimated to be 12.3 per 100,000 people in the United Kingdom (Evans et al., 2013). Random assortment of couples would lead to roughly 1 couple per 67 million couples in which both parents are heterozygous carriers, corresponding to roughly 3 couples in the United States and Europe combined. As noted earlier in the chapter, to meet the criteria identified by the Commission for initial clinical uses of HHGE would also require having a genome editing methodology capable of reducing the number of trinucleotide repeats to a level typical of unaffected individuals.
Mutations in the gene presenilin 1 (PSEN1) cause early onset Alzheimer’s disease. Although PSEN1 mutations are the most common cause of early onset inherited Alzheimer’s disease, determining the frequency of PSEN1 mutations in a population is complicated by the fact
that multiple possible mutations (not only in PSEN1 but also in the genes PSEN2 or APP) can cause this disease and by the fact that there are also later onset forms of Alzheimer’s and dementia. It has been estimated that up to 1 percent of Alzheimer’s cases arise from gene mutations in PSEN1, PSEN2, and APP;11 other estimates have indicated that 50,000 to 250,000 people in the United States have early onset Alzheimer’s disease, occurring prior to age 65.12 Estimates of the frequency of PSEN1 mutations in global populations are not readily available. Cases in which both prospective parents carry the mutation may be more common where there are higher rates of consanguineous marriage.
Mutations in the gene APC cause the disease familial adenomatous polyposis, which results in the development of colon cancer by middle age as well as increased risk of cancer in other organs. Familial adenomatous polyposis has been reported to occur in 1 in 7,000 to 1 in 22,000 people.13
Although both parents would need to carry alleles for a serious dominant disease to meet the circumstances identified by the Commission for potential initial uses of HHGE, it may not be necessary for parents to carry alleles for the same disease. It could be possible that each parent is heterozygous for a different dominant disease. By probability, the embryos such a couple could produce would still have only a 25 percent chance of being unaffected by a serious disease. However, the use of HHGE in such a circumstance would entail decisions about whether to attempt genome editing of more than one disease-causing allele or which disease to target through the editing process.
These examples help illustrate that circumstances in this very small subset of Category B are likely to be rare. However, such cases are expected to exist. Dominant mutations may be found at significant frequency in founder populations, and union between individuals with the same or different mutations is not rare. In addition, several PGT clinics in the United States and Western Europe indicated that they have seen patients whose embryos would have a low chance of being unaffected by a genetic disease (personal communications). Although detailed data were not available to the Commission, a preliminary estimate was up to 1 such couple per year per clinic, as compared to 50–100 couples seen for circumstances in Category B where embryos would have a 50 percent chance of inheriting a disease-causing genotype.
Considerations After Initial Human Uses
Should first-in-human uses take place and appear to be successful, without raising concerns about safety and efficacy, it may become appropriate to consider the use of HHGE in additional circumstances in Category B. Such a decision could enable evidence to be obtained on whether or not HHGE followed by PGT provides an improved option compared to PGT alone for prospective parents wishing to prevent transmission of a serious monogenic disease. However, this would require that a controlled clinical evaluation (randomized control trial) be designed to compare the success rates of these two types of interventions (PGT alone in one arm versus HHGE with PGT in the other). Such evidence would answer questions that have been raised about whether, in particular genetic settings, HHGE can increase the numbers of high-quality embryos available for transfer for couples in which some embryos will inherit disease-causing genotypes, and the results would inform discussions on future clinical practice. The numbers of couples who would take part in any initial clinical uses of HHGE to evaluate safety and efficacy is expected to be too small to design and recruit participants for such an evaluation. Moreover, the comparison would depend on the genetic setting (specifically, the expected proportion of unaffected embryos). Conducting such evaluation would thus require the inclusion of many additional participants in Category B. Evaluating the results of any initial human uses and making decisions on whether to consider any further uses of HHGE would require national and international processes described in Chapter 5.
Fundamental laboratory research (not undertaken with the clinical aim of establishing a pregnancy) related to genome editing of human gametes, zygotes, and embryos is itself important.
To Better Understand Human Embryo Development
The understanding of human embryo development is an important area of research. Genome editing has already provided major new insights into preimplantation human development. Such research on human embryos, while raising ethical issues of great importance, is scientifically essential because there are considerable differences between species. Such studies will lead to better understanding of the reasons for the limited success of IVF for some prospective parents and may well help our understanding of female infertility and miscarriage. Research using genome editing in human embryos will also give important insight into the effects of maternal aging on human embryo development, an area of increasing interest with a growing
number of women choosing to delay pregnancy. It will also shed light on mechanisms of DNA repair that operate specifically in the early embryo, a process that will inevitably need to be controlled in order for the outcomes of genome editing to be completely predictable and precise. Finally, it will help in understanding the role that key genes play in specifying cell fate in the human embryo, which may have profound implications for our ability to culture and manipulate human stem cells for applications in regenerative medicine. To perform such research to the highest standards, in which a particular embryonic phenotype can be attributed to a specific genetic event, researchers will require genome editing protocols of the highest efficiency and specificity.
To Improve on Assisted Reproductive Technologies
Fundamental research to improve the general ability to precisely edit the human genome, control on-target events, avoid mosaicism, and generate no off-target effects could improve the utility of HHGE in an assisted reproduction context. If HHGE could be performed very safely and at extremely high efficiency, it could be possible to use it to increase the number of embryos not carrying the disease-causing genotype available to prospective parents undergoing PGT, which might allow expanding use broadly in Category B.
It is not possible to define a responsible translation pathway for all possible uses of HHGE, because the benefits and risks depend on particular circumstances, including the severity of the disease, the genetic situation of the couple, the mode of inheritance of the disease, the nature of the proposed sequence change, and the availability of alternatives. Given the uncertainties inherent in a new technology like HHGE, clinical evaluation should proceed incrementally, cautiously, and with humility, initially focusing only on those potential uses for which available knowledge has established an evidence-base and for which the balance of potential benefit and potential risk is carefully evaluated to ensure a high benefit-to-harm ratio.
To achieve this balance, the Commission concludes that any initial uses of clinical HHGE must meet all four criteria identified in this chapter. At present, it is only possible to define a responsible clinical translational path for applications of HHGE that fall into Category A or, possibly, a very small subset of Category B. For all other circumstances, additional considerations and lack of knowledge make it impossible today to properly evaluate the balance of risks and benefits, and the Commission is not currently able to describe a responsible translational pathway for clinical use.
Recommendation 3: It is not possible to define a responsible translational pathway applicable across all possible uses of heritable human genome editing (HHGE) because the uses, circumstances, and considerations differ widely, as do the advances in fundamental knowledge that would be needed before different types of uses could be considered feasible.
Clinical use of HHGE should proceed incrementally. At all times, there should be clear thresholds on permitted uses, based on whether a responsible translational pathway can be and has been clearly defined for evaluating the safety and efficacy of the use, and whether a country has decided to permit the use.
Recommendation 4: Initial uses of heritable human genome editing (HHGE), should a country decide to permit them, should be limited to circumstances that meet all of the following criteria:
- the use of HHGE is limited to serious monogenic diseases; the Commission defines a serious monogenic disease as one that causes severe morbidity or premature death;
- the use of HHGE is limited to changing a pathogenic genetic variant known to be responsible for the serious monogenic disease to a sequence that is common in the relevant population and that is known not to be disease-causing;
- no embryos without the disease-causing genotype will be subjected to the process of genome editing and transfer, to ensure that no individuals resulting from edited embryos were exposed to risks of HHGE without any potential benefit; and
- the use of HHGE is limited to situations in which prospective parents (i) have no option for having a genetically-related child that does not have the serious monogenic disease, because none of their embryos would be genetically unaffected in the absence of genome editing; or (ii) have extremely poor options, because the expected proportion of unaffected embryos would be unusually low, which the Commission defines as 25 percent or less, and have attempted at least one cycle of preimplantation genetic testing without success.
Chapter 4 sets out the elements that would be required for a responsible translational pathway toward initial uses of HHGE, in the event a country were to permit such uses.