Based on historical precedents preceding even the first production of hESCs, it seems certain that one of the primary concerns of the FDA regarding clinical use of hiPSC-based biologics is going to be safety, particularly as it relates to how the methods used to produce hiPSCs impact safety. In this regard, it is important to distinguish between two main types of hiPSC-based therapeutic products: the use of hiPSCs themselves for transplantation versus employing products made from hiPSCs, such as differentiated cells or proteins produced by the hiPSCs. The former approach is widely viewed as greatly more challenging given the teratoma-forming activity of hiPSCs. To date, the latter approach has been the one primarily pursued in the form of differentiated derivatives based on hESCs. For example, ACT has two ongoing combined Phase I/II clinical trials for treating macular dystrophy using differentiated RPE cells, and Geron (studies now halted) used hESC-derived oligodendrocyte precursors to treat spinal cord injuries. The experiences of ACT and Geron are instructive for clinical applications of hiPSCs in terms of the requirements for preclinical data prior to approving Investigational New Drugs (INDs) and also for early-phase clinical trials based on hESCs. One can imagine in the future similar kinds of studies based on hiPSC-derivative differentiated cells as well. Another important distinction related to hiPSC therapies in development is their autologous and allogeneic use. While in principle hiPSCs are from one perspective inherently most ideal for autologous therapies, in practice hiPSCs could be used for both autologous and allogeneic therapies. Indeed, it is even possible that the first clinical use of hiPSC-derived therapies could be allogeneic in nature.
For both ACT and Geron, the FDA has been appropriately fastidious in terms of the requirements for preclinical data prior to approving INDs and also for early-phase clinical trials based on hESCs. An informative example of this is the nearly year-long delay for Geron because of what turned out to be cysts at the sites of injection. While the cysts appeared benign and the trial was ultimately allowed to proceed, I believe the caution was in order given the potential risks of uncontrolled cellular behavior in transplant recipients.
Novocell (now Viacyte) is a biotechnology company currently conducting preclinical studies for an encapsulated hESC-based pancreatic β-cell progenitor (Pro-Islet™
) therapy for Type 1 diabetes. In preclinical studies, the therapy effectively induced insulin-sensitivity and stable blood sugar in mice [6
]; however, Viacyte’s diabetes therapy still remains in the preclinical phases of development. By contrast, ACT is now well into its clinical trials and recently reported no adverse outcomes in the first few patients in its trials, providing encouraging early results in terms of safety [7
]. Why is ACT moving more rapidly than Viacyte? At least one difference is intrinsic to the target pathology, with Type 1 diabetes being an autoimmune disease, providing Viacyte with some unique challenges such as recipient rejection of a cellular therapy. While hESC-based, the case of Viacyte is instructive for hiPSCs as well, since the use of hiPSC therapies for diabetes would also face similar challenges. While hESC-based therapies have faced and continue to face significant regulatory hurdles, as illustrated by the above examples, it is probable that the FDA will need to be even stricter with hiPSC-based therapies given the extra steps required to produce hiPSCs relative to the derivation of hESCs.
I predict that the FDA will view hiPSCs in the context of their experiences with hESC-based products, with many of the same issues to be addressed for both. The most prominent concern is safety in the form of either tumorigenesis (e.g., teratoma) or production of undesired tissues, given the pluripotency of hiPSCs. However, hiPSCs face unique challenges as well, including the methods used to produce them and the predicted instability. In this regard, it is well documented that hESCs tend to accumulate mutations over time in culture, including many related to oncogenes [8
], and the same was predicted to be true for hiPSCs. One type of genomic change – copy number variations (CNVs) – occurs when individual cells have gain or loss of specific genes. For example, cancer cells often have CNVs reflecting gains of oncogenes and losses of tumor suppressor genes. Interestingly, the number of CNVs in hiPSCs, which was higher than in hESCs, was observed to fall over time in culture [16
], suggesting many CNVs were somehow selected against. Importantly, the levels of CNVs in hiPSCs were generally equal to or higher than in hESCs, indicating that in this regard hiPSCs do not have an advantage over hESCs. Furthermore, the drift of CNV numbers in hiPSCs at the same time also demonstrates their inherent changeability in culture, an undesirable property from a therapeutic standpoint. Of particular concern is the fact that CNVs in hiPSCs and hESCs both frequently involved gains in copy numbers of oncogenes. Methods to enhance iPSC stability during production and culture have been suggested, including addition of factors during reprogramming that promote genome stability [4
The epigenetic state or ‘epigenome’ of hiPSCs is remarkably similar to hESCs, but also contains imperfections and ‘memories’ [17
] of unknown biological function. During cellular reprogramming, the somatic cell epigenome is reprogrammed as well. However, a variety of differences in the epigenome are apparent in hiPSCs as compared with hESCs. This incomplete epigenomic reprogramming includes regions of persistent DNA methylation and histone modifications that are distinct from those observed in hESCs. The inherently more dynamic state of the epigenome versus the genome makes it reasonable to predict that the hiPSC epigenome will prove changeable during large-scale culturing as well. While it is unclear whether in the future the FDA will require epigenomic data as part of the approval process of hiPSC-based therapies, such data could prove highly informative in the evaluation of biologics and potentially in the regulatory decision-making processes. Given these factors, how might the epigenomic state of hiPSCs or their derivatives be most effectively assessed? Current state-of-the-art technology for globally assessing the epigenome is chromatin immunoprecipitation sequencing (ChIP-Seq) using antibodies for specific histones or methylated DNA. As sequencing technology advances, it is reasonable to expect that ChIP-Seq will become faster and cheaper, making it potentially more applicable to preclinical screening of stem cell lines (reviewed in [20
]). Given that epigenomic reprogramming is intrinsic to the production of hiPSCs and that hiPSCs are not identical to hESCs, I recommend some level of epigenetic assessment be incorporated into the clinical hiPSC validation process. Other potential levels of validation that are predicted to be informative include proteomics and metabolomics profiling [21
], as the proteomes and metabolomes of iPSCs are not identical to hESCs.
An additional challenge regarding hiPSC safety is the relative dearth of safety-related studies that have strong clinical relevance [4
]. With few exceptions, the tumorigenicity of hiPSC lines has been evaluated using classic so-called ‘teratoma assays’. In these assays, hiPSCs are most often injected subcutaneously into immunodeficient mouse models. While such assays are effective tools for evaluating the pluripotency of stem cell lines, teratoma assays have a number of weaknesses. The lack of standardization in how they are performed [23
] and their dissimilarity to the context of how hiPSCs would actually be transplanted in a true human clinical setting are problematic. For example, most human patients would be immunocompetent and injection of cells subcutaneously may have little relevance to any potential human tissue into which hiPSCs may be transplanted in the clinical setting. What are potential alternatives to teratoma assays? One intriguing approach, PluriTest™
(Scripps Research Institute), relies on gene expression profiling that accurately assesses pluripotency without the use of teratoma assays [24
]. It may be possible to glean more precise and useful preclinical information on pluripotency signatures through genomics approaches as well, as discussed below.
There currently is no regulatory mandate that hiPSCs should be produced using non-genetic methods. If genetic methods are used to make hiPSCs intended for clinical use, as a consequence, the potential hiPSC-based therapy in question would be both a cellular and gene therapy product. What this means is that any such genetically based hiPSC therapy would be subject to significantly more challenging regulatory hurdles than, for example, hESCs or hiPSCs made without genetic modification. Thus, while the FDA has not publicly banned or even advised against the future clinical use of genetically produced hiPSCs as therapies outright, scientists wanting to translate such therapies to the bedside may be somewhat daunted by the extra regulatory hurdles. However, the use of genetically modified cells has precedent and may prove valuable. For example, ReNeuron’s fetal brain-derived cellular product RN001, approved by UK regulators for clinical trials in stroke patients, was genetically modified, but it is notable that the company is performing the trial outside the USA after a prolonged period of FDA review.
The use of genetic methods, such as integrating viruses to make hiPSCs, present potentially more intrinsic problems in the form of cells that may be predisposed to be less safe due to the effects of transgenes [25
]. For example, the use of Myc
genes has long been a concern in a murine context due to its demonstrated ability to cause tumors in hiPSC-derived mice [26
], but whether endogenous or exogenous, Myc proteins fulfill essential roles in both embryonic stem cells and hiPSCs [27
] and pose potential risks. Nonetheless, the search for completely nongenetic, reasonably efficient hiPSC production technologies may not be as entirely essential from a regulatory perspective as is widely assumed in the field. In addition, it may be possible to replace c-Myc
with the potentially safer L-Myc
or with mutant forms of Myc
that are reportedly still functional for reprogramming but far less oncogenic [29
]. Alternatively, Myc
may be replaced with other genes that appear safer, such as UTF1
, which has already been shown to effectively substitute for Myc
], or others, such as Lin28
. It is notable that the field has greatly advanced in the last few years, as reflected in the declining number of reprogramming factors required to produce hiPSCs, pointing to Oct4
being the most essential. The use of small-molecule drugs during reprogramming has greatly added to the efficiency of such methods using fewer factors. More broadly, the field is shifting towards the use of nongenetic approaches to hiPSC production, such as episomal vectors and viruses such as adenovirus, which may ultimately facilitate most future hiPSCs being made using non-genetic methods [31
]. However, at this time, genetic approaches remain by far the most efficient and most widely used. Importantly, the issue of genetic versus nongenetic hiPSC production becomes largely moot in the case of hiPSC-produced drugs used for therapies.