4A Translational Pathway to Limited and Controlled Clinical Applications of Heritable Human Genome Editing

Controlling On-Target Events

The inability to control events at the genomic target site constitutes a major limitation for HHGE. The majority of disease-causing mutations would require introducing the non-disease-causing sequence by copying from a provided template or from a non-disease-causing gene copy located on the homologous chromosome. Based on limited experience, this process of HDR is not efficient in human zygotes following a double-strand break in the DNA. In zygotes and in other cell types, the more common outcome of making a double-strand break is the introduction of sequence insertions and deletions (indels) via NHEJ. The NHEJ process could result in replacing one mutation with another, the consequences of which cannot be predicted or controlled. Such products would be deleterious in almost all instances; therefore, the ratio of DNA repair by HDR to NHEJ must be increased to achieve the desired outcome with high probability. As noted in Chapter 2, both base editing and prime editing largely avoid the risks associated with making and repairing a double-strand break; and both (particularly base editing) have shown promise in embryos.

The goal of HHGE would be to generate embryos that carry only a common, non-disease-causing sequence at both alleles of a gene. Creating one non-disease-causing allele would still be effective in the case of a recessive condition, while restoring both would also eliminate carrier status. A corollary of this goal is that alteration of a pre-existing non-disease-causing allele should be avoided. Zygotes that require genome editing must be identified prior to treatment. The latter may be possible through biopsy and testing of the first and second polar bodies (see Chapter 2) or future development of efficient genome editing methodologies for multi-cellular embryos that have already been genotyped.

Minimizing Off-Target Events

The ability to reduce the frequency of unintended genetic changes and to detect such changes when they occur has progressed significantly in recent years. For CRISPR-based genome editing, testing of various guide ribonucleic acids (gRNAs) for a particular target and making modifications to both the gRNA and the Cas protein have improved specificity. Similar advances have been made for the zinc-finger nuclease (ZFN) and transcription activator–like effector nuclease (TALEN) platforms. However, there are still challenges for detecting unintended sequence changes with high confidence in embryos. Analysis of off-target events arising from genome editing can be done by whole-genome DNA sequencing; however, current whole-genome sequencing (WGS) methods are not adequate for the accurate analysis of the small amount of genetic material that can be safely extracted from blastocyst-stage embryos intended for transfer to the uterus. In addition, WGS may not capture the full range of alterations that can occur. These could include large insertions and deletions or even whole or partial chromosome losses, which are difficult to detect with WGS or with standard polymerase chain reaction–based procedures.

Minimizing Mosaicism

Preventing mosaicism requires the ability to make the desired on-target modification with very high efficiency either in the one-cell zygote with restriction of editing activity to that stage or in all cells of embryos comprised of two or more cells. If genome editing continues beyond the first cell division, different cells in an embryo may carry different sequence changes at the intended target or at off-target sites. The effect of such mosaicism is difficult to predict, but it may pose serious risks by either failing to prevent disease due to target tissues having an insufficient number of appropriately edited cells or by introducing undesired mutations—particularly large copy number variants—at the target locus or elsewhere in a fraction of cells that could result in diseases related or unrelated to the targeted disease. Mosaicism poses particular challenges to verification. For an embryo destined for transfer, only a few trophectoderm cells can be safely removed from a blastocyst for molecular analysis. No current method can determine whether all cells of an embryo intended for uterine transfer carry exactly the same edits; it is even difficult to envision one that would. This means that preclinical research must establish procedures that only very rarely lead to mosaic embryos.

Evaluating the Physiological Effects of Genome Editing of Disease Alleles

Research on the short-term and long-term physiological and functional outcomes of editing disease alleles is needed to verify that a given intended edit is sufficient to prevent the disease phenotype and to provide reassurance that significant, unanticipated health effects would be unlikely to result from the genome editing process. Useful information may be obtained from human somatic editing of the same disease allele and potentially from the use of germline editing in other mammals—for example, to alter the animal-equivalent version of the human allele in cases where the human disease phenotype is accurately reproduced.

Need for Extensive Research in Cultured Human Cells and in Zygotes of Model Organisms

Each combination of a specific target gene and editing reagents would need to be evaluated, since each combination will present unique potential for on- and off-target events. To justify the design of the editing reagents proposed for potential clinical use, preclinical research in cultured human somatic cells from the prospective parents and in model organisms must include the following steps.

Assessment of Parental Genomes

Requirement: Obtain whole-genome sequences of the prospective parents using best practice protocols for investigating genetic disorders. Identify the exact sequence of the target mutation and surrounding genomic region. For a given combination of target and editing reagents, assess potential off-target sites based on these genomes.

Context: WGS is routinely used to identify new (de novo) mutations in offspring, as well as to establish the specific disease-causing genetic variant that parents with a history of genetic disease are at risk of passing on. Examples of best practice protocols include the Deciphering Developmental Disorders study in the United Kingdom.¹

Testing of Genome Editing Reagents in Cultured Parental Cells

Requirement: The following assessments need to be undertaken:

• For assessing on-target efficiency, test the editing reagents using cells from the parent(s) with the disease-causing mutation.
• For identifying sites at risk of off-target editing, test in cells from both parents.
• If a non-disease-causing allele is present in the genome of either parent, also test for any potential undesired editing of this allele.

Context: Testing in parental cells is important to allow for possible effects of genetic background on on-target and off-target outcomes. The information obtained from these assessments should be used to refine the editing reagents for efficacy at the intended target and to assess and minimize off-target mutagenesis. The cumulative frequency of off-target mutagenesis should not be significantly higher than the expected de novo mutation frequency. The cells could be primary cells from each parent, induced pluripotent stem cells (iPSCs), or embryonic stem cells (ES) derived from the parents by nuclear transfer. Because the adaptation to culture and the induction of pluripotency can both lead to the accumulation of novel mutations, testing in several independently derived lines is advisable.

Testing of Editing Reagents in Embryos of Model Organisms

Requirement: Test the efficiency of modifying the comparable target sequence in zygotes from a mammalian model organism. Use models incorporating humanized sequences—at least the sequence to be modified at the target and surrounding regions recognized by the editing reagent.

Context: Genome editing in mammalian zygotes differs from editing in somatic cells of the same species. Since some embryo-specific characteristics are likely shared among species, testing the editing reagents in zygotes of mammalian model organisms allows the characterization of the types of editing outcomes and the development of procedures to prevent and assess mosaicism. Relevant contextual factors in mammalian embryos include the methods of delivering the editing reagents, which are different in zygotes compared with cells in culture, and the DNA repair mechanisms active in zygotes compared with adult cells. While processes in human zygotes may differ from those in other mammalian zygotes, information obtained from such experiments can provide guidance for further testing in the human context.

Testing in mammalian zygotes is essential for refining the editing system prior to preclinical testing in human embryos. This testing is not designed primarily to evaluate phenotypic effects of editing the target sequence in the model organism. Animals such as mice can be used to generate sequences equivalent to the human disease-causing allele, so-called humanized alleles (Zhu et al., 2019). Evidence showing that the humanized disease-causing allele can be edited to the non-disease-causing allele would be essential to have prior to the earliest clinical uses of HHGE. If the humanized sequence cannot be produced in a mammalian model organism for some reason, then this disease allele should not be selected for an initial application of HHGE.

Preclinical Testing in Human Embryos

Preclinical testing in human zygotes must be undertaken to demonstrate that the genome-editing methodology proposed for clinical use provides high levels of efficiency, specificity, and safety. No other cell type can substitute for this stage of preclinical evidence. The preclinical testing of human embryos is conducted in a laboratory, and the embryos are never used to establish a pregnancy.

Only a limited number of human zygotes are available for experimental purposes, and the Commission recognizes that many jurisdictions do not permit the creation of human embryos for research. Nonetheless, thorough validation of the genome editing process prior to clinical use would require data from human embryos. To minimize generation of embryos specifically for experimentation, zygotes created through ARTs but not used by a couple to establish a pregnancy may be donated for use in laboratory research. While such zygotes would likely lack the specific disease-causing allele(s) being targeted for the proposed use of HHGE, testing them would generate information about potential off-target editing and could provide valuable guidance regarding zygote-specific processes involved in on-target editing. This option is limited by the fact that these stored zygotes will likely be at the G2 cell cycle stage—that is, later than would be subjected to genome editing—but may still be useful for approaches that address this stage or even ones in development for two-cell embryos. In addition, most in vitro fertilization (IVF) embryos are currently stored at even later stages and may not be useful at all.

For initial uses of HHGE, there will be certain types of information that could only be obtained following human clinical use. As a result, the standards for preclinical testing must be very high, and human zygotes for preclinical assessment of the editing methodology should be obtained that contain the disease-causing mutation. The efficacy of the editing reagents on disease-causing alleles carried by the prospective male parent can be tested in zygotes produced using his sperm to fertilize donated oocytes. In the case of disease-causing alleles transmitted by the female parent, care must be taken to avoid subjecting her to multiple rounds of hormonal stimulation and oocyte collection, with their attendant risks. If she is not at further risk from the IVF process and not of advanced age, a single round of stimulation and collection might be appropriate for the purposes of generating embryos for testing. A reasonable alternative would be to recruit a sperm donor who carries the same disease-causing mutation(s) and to use his sperm in conjunction with donated oocytes to produce zygotes for testing.

After a substantial knowledge base has been gained through experimentation on human embryos, it might be possible to identify alternative cell-types that reliably allow accurate prediction of the effects of genome editing in human zygotes. In such circumstances, it might become acceptable to use these cells as a surrogate for the preclinical tests involving human embryos required in this pathway. Rigorous scientific assessment of such models and their ability to substitute for evaluation in human embryos would be critical before using such alternative cell systems as the only source of preclinical evidence for HHGE. Over time, extensive preclinical testing of the ability of a particular editing methodology to correct a variety of targeted alleles might also become considered sufficient to conclude that it was not necessary to test the correction of each specific allele in preclinical human zygotes. Ongoing and independent scientific and technical reviews to assess knowledge gained and to consider whether it may be reasonable to make changes to preclinical standards required for the earliest human uses would be crucial (see Chapter 5 for a discussion of such oversight issues).

As noted above, preclinical testing can be complicated by the particular genetic circumstances of each couple. For example, when one or both prospective parents is a compound heterozygote for a serious genetic disease, more than one disease-causing allele will be present for that disease. This could pose a challenge to the design of gRNAs as they may only be able to edit one disease-causing allele and not others. Preclinical testing will need to examine the effects of the editing reagents on both the targeted allele and any other alleles present. In such circumstances, PGT on any edited embryo intended for uterine transfer will have to determine that at least one disease-causing allele has been changed to a non-disease-causing allele that is common in that population and that the other allele is unaffected.

Preclinical testing in human embryos must include the following steps.

Characterizing Editing at the Target Site

Requirement: The efficiency of the intended edit must be very high when measured in a cohort of treated human embryos. There must be no other sequence changes induced at the target, including insertions and deletions (indels).

Context: The goal of this testing would be to guarantee that the genome-editing methodology produced sufficient numbers of high clinical grade embryos with the desired edit before moving further toward to any clinical use. For a dominant disease, both zygotic alleles would need to exhibit a non-disease-causing sequence. For a recessive disease, although the restoration of a non-disease-causing sequence in one allele would prevent the disease, the editing frequency must be high enough that a high proportion of the available embryos are so modified. This testing could be done at any multicellular stage and must include testing for large deletions, chromosome loss, and other rearrangements.

Characterizing Any Off-Target Editing

Requirement: Compare parental genomes with whole-genome sequences obtained from the edited embryos or ES cells derived from these. Targeted sequencing should also be done for any particular off-target sites identified in preclinical research. There must be no detectable editing-induced off-target sequence changes. The incidence of de novo mutations, determined in conjunction with the sequence of the biological parents, must be in the range observed for unedited embryos, with no increase in the occurrence of single-nucleotide polymorphisms, indels, copy number variants, or chromosome rearrangements.

Context: Testing would be done at the blastocyst stage to provide sufficient DNA for this analysis and because this is the stage at which embryos would be transferred in the case of actual clinical use.

Identifying off-target sites from analysis of the cultured parental cells (see the first step under “Need for Extensive Research in Cultured Human Cells and in Zygotes of Model Organisms,” above) would provide an initial indication of high-risk off-target sites in the embryo, allowing attention at this stage to focus on regions where off-target editing has previously been observed or might be anticipated to occur.

Characterizing Any Mosaicism

Requirement: All cells of the embryo must have the same on-target sequence (i.e., no mosaicism) as shown by analysis of multiple individual cells. The sequences of the intended target and any high-risk off-target sites should be determined for each individual cell, or as many as feasible, in the early-stage embryo.

Context: If not all cells have been successfully edited, it is possible that the target organ(s) in the offspring, and/or later in the adult, will not be completely disease-free.

Characterizing Embryo Development

Requirement: The genome-edited embryos must proceed through normal development in vitro to the blastocyst stage, meeting milestones with comparable efficiency to unedited embryos. Cellular and molecular features of genome-edited embryos should be comparable to unedited embryo controls and have aneuploidy rates no higher than expected based on standard ART procedures.

Context: The goal of such testing is to ensure that the genome editing does not negatively affect normal embryo development. Developmental characterization in genome-edited embryos would be compared to expected embryo development based on what is known from the use of unedited embryos in IVF. This assessment could be continued up to the 14-day limit currently permitted for human embryo culture in many countries.

Examples of best practice protocols include those used by the Newcastle Fertility Centre at the International Life Science Centre and others in the preclinical evaluation of human embryos that had undergone mitochondrial replacement techniques (MRT). In collaboration with the Crick Institute, investigators at Newcastle analyzed the cell lineages that were present in blastocyst-stage human embryos that had undergone MRT, to ensure that all expected cell lineages were present. They also performed single-cell transcriptome analysis to check for the expected patterns of gene expression (Hyslop et al., 2016).

Only if all of these preclinical requirements are met and validated by independent expert opinion should the use of edited embryos in a clinical setting be contemplated.

Additional Consideration for Genome Editing in Which Not All Embryos Would Inherit the Disease-causing Mutation

Requirement: Development of a genome editing methodology capable of safely and efficiently editing an eight-cell or later embryo may be required.

Context: If all of the embryos that can be produced by the prospective parents will inherit the disease-causing mutation, genome editing could be undertaken around the time of fertilization. If only some of the embryos will carry the disease-causing mutation, identifying which are affected would be necessary prior to editing. For genetic circumstances in which only the maternal genetic contribution needs to be known, identifying the oocyte genotype through polar body analysis may be sufficient. In other circumstances, in which both maternal and paternal genetic contributions need to be known, an embryo biopsy would be required. For any such envisioned uses, the preclinical development and testing of a genome editing methodology capable of safely and effectively editing embryos post-genotyping would be required.

Monitor a Resulting Pregnancy

Requirement: Careful monitoring of a resulting pregnancy with a genome-edited embryo is strongly recommended.

Context: Following transfer of a genome-edited embryo to establish a pregnancy, prenatal monitoring is crucial to detect any fetal abnormalities or other issues arising during the pregnancy. Should prenatal testing identify genetic or physical anomalies, counseling for the parents will be important. Such prenatal monitoring would be strongly recommended but is the choice of the mother.

Undertake Longer-term Monitoring and Follow-up

Requirement: Longer-term monitoring and follow-up of a child born following HHGE is essential, and should include:

• obtaining consent from the parents, and later from the child, for monitoring immediately after birth and at specified intervals thereafter extending into adulthood, which must be done by competent professionals and include both physical and psychological aspects; and
• using assessment tools that have been validated and standardized in an international context and, if appropriate, that have versions available across the lifespan.

Context: It is important for the health of the individual born as a result of HHGE, as well as any children that they have, that such individuals continue to be assessed for adverse genetic or health outcomes. If adverse outcomes are identified, the individual concerned should be informed of them if they so choose and offered genetic counseling.

Make Information on Decisions to Permit the Clinical Evaluation of Heritable Human Genome Editing Publicly Available

Requirement: Each country would need to make public the details of any approved applications to clinically evaluate HHGE. Information that would need to be made available includes the genetic condition for which HHGE had been allowed, the associated laboratory procedures that would be used, and the national bodies that would be providing oversight.

Context: Making such information publicly accessible would be important for ensuring transparency about any potential uses of HHGE being contemplated, the evidence base on which decisions had been made, and oversight responsibilities. However, information made available would need to protect family identity.

Evaluate Information to Inform Future Decisions about Heritable Human Genome Editing

Requirement: It would be vital to publish in peer-reviewed scientific journals the procedures involved and outcomes of any clinical evaluation of HHGE.

Context: Such information would contribute to ongoing national and international discussions on the safety and efficacy of HHGE. In conjunction with further extensive societal engagement, such information would also contribute to any decisions about whether to consider the clinical evaluation of HHGE for other uses in Categories A and B, according to the translational pathway identified in this report, whether or how to modify any of the preclinical or clinical requirements laid out in this translational pathway, or potentially whether to consider evaluating uses that would fall into other categories of potential uses described in Chapter 3.