Discussion
To gain insight into the non-clinical development program for cell-based products we analysed the non-clinical data package of 86 different cell-based products based on scientific advices provided by the EMA between January 2013 and June 2018. A 5-year time period was chosen to ensure that our data contained a sufficient number and diversity of cell-based therapies to allow for a meaningful analysis of the non-clinical development program. In comparison, a previous analysis of the non-clinical program in EMA advices on ATMPs contained 54 ATMPs of which thirteen were somatic cell therapy and eight were tissue engineered products[22].Within the 86 products analysed, a variety of autologous or allogenic cell-based products with and without genetically modifications were included, however the class of products containing genetically modified allogenic cells was only represented by six products.
A non-clinical development program consists of in vitro andin vivo studies designed to provide a clinical treatment rationale (PoC) and to gain insight into the safety profile of a medicinal product. For cell-based therapies translation of animal data towards the human situation has significant limitations. When testing a human cell-based product in an animal (heterologous model) limitations include possible differences in cell size or cell metabolic rate [23], immunogenicity (xeno-reactions can occur extremely rapid and vigorous) [24] and potential species specificities in cell-cell or cell-environment interactions [25]. These limitations all need to be considered when interpreting study results. A homologous model in which the animal equivalent of the human cell-based product is tested, could be considered as an alternative in vivo testing strategy. However, interpretation of results from such a model may be complicated by differences in manufacturing processes of the animal equivalent, deviation of animal response from the human response or absence of an appropriate animal equivalent of the human cells[25].
Interestingly, while the translation of animal data to the human situation has significant limitations for cell-based products, only for six products animal studies were not performed or proposed by the sponsors nor requested by the regulators. These products contained cell types, which have been studied in humans for a relatively long time, such as dendritic cells [26], antigen-specific T-cells [27] and mesenchymal stromal cells (MSC) [28]. Apparently, for these products it was considered that additional animal studies would not provide new insights on these types of products. As the clinical experience with cell-based products is increasing, the knowledge on the (potential) effects of the studied cell types is also increasing. This may result into a reduction or even omission for the need of in vivo animal studies for certain types of cell-based therapies.
Still for most products the sponsors considered data from animal studies of value for the development of their product. This was particularly the case for studies on PoC as in vivo pharmacology studies were performed for almost all the products. Also safety was studied in vivo in animals for the majority of the cell-based products, albeit more often in combination with other endpoints e.g. by also addressing the cellular biodistribution or by including the safety endpoints in pharmacology studies. Thus, it seems that the focus of the non-clinical development program for cell-based products is more on pharmacology/PoC than on safety. This is different from what we see for more ‘conventional’ medicinal products, where safety is an important component of the non-clinical in vivo studies and almost exclusively evaluated in dedicated toxicity studies.
The fact that pharmacology is most often studied in a dedicated study could be explained by a difference in timing of the studies, as non-clinical data are mostly intended to support the rationale for further development before safety studies are performed to ensure the safety of the first clinical trial subjects. Pharmacology studies do not require GLP compliance, in contrast to safety studies. This may have contributed to the higher number of products with pharmacology studies. Notably, although GLP-compliance is formally required for non-clinical safety studies for human medicinal products, for cell-based therapies it has been acknowledged that GLP compliance of in vivo safety studies may not always be feasible [29].
In our analysis we particularly focussed on the need for and the type of biodistribution and/or tumourigenicity studies since these aspects are very difficult to study in humans. For more conventional medicinal products, the kinetics of the product and its safety profile is studied in animals. Interestingly, in our analysis in vivo animal studies on biodistribution and/or tumourigenicity were not considered necessary by both sponsors and regulators for approximately one third of the cell-based products irrespective of product type. This observation suggests that for cell-based products a more tailored non-clinical development program is necessary. Thereby sponsors should focus on the need to better characterise the properties of the cell-based product and consider the (im)possibilities of studying a products behaviour in an animal model.
Even though a considerable number of cell-based product in our database are lacking an in vivo study on biodistribution, still this type of study was performed for two-thirds of the products and considered to be of value for understanding the PoC and to gain insight into the safety profile. For biodistribution most often a heterologous animal model was used. This is remarkable as cell migration is expected to be dependent on chemotactic signals and cell-cell interactions which might be species-specific and may thus better be studied in a homologous model. Next to distribution to target and non-target tissue, also persistence of the cells was (proposed to be) evaluated for approximately two third of the products, suggesting that sponsors already acknowledge persistence as an important biodistribution endpoint to be investigated. Detection of administered cells in target and non-target tissues was most commonly done by tissue sampling followed by (semi)quantitative detection of these cells with various sensitive techniques. Notably, more sophisticated techniques based on PET, SPECT, and MRI imaging are being developed, which can trace living cellsin vivo [14,20], allowing for serial sampling and whole-body scanning. However, these techniques were rarely used to study the biodistribution of the products in our database. Possibly these techniques were still not sufficiently developed at the time of the design of the non-clinical studies for our analysed products. It is also possible that these techniques were not (yet) attractive due to high costs, limited access and/or the perception that these techniques are too novel and have not yet proven their value to ensure their acceptability by regulators. Notably, these newer techniques of cell tracking may also become suitable for the evaluation of biodistribution of cellular therapeutics in humans, thus in the clinical setting. Clinical biodistribution studies would be of much more relevance as the therapeutic product can be tested directly in the target ‘species’. The hope is that these more advanced methods analysing distribution of cellular therapies, will be implemented in due time and that they could possibly obviate the need for extensive evaluation of in vivobiodistribution of human cell-based products in animals.
As tumourigenicity is a concern associated with cell-based products, it is not surprising that for most products experimental data was used to address the potential for tumourigenicity. For half of the products this was planned or performed to be addressed by means of in vivoanimal studies. For four-fifths of these products, in vitrostudies supplemented the tumourigenicity evaluation. One-fifth thus leaned on in vivo studies only. The design of these in vivo studies appeared rather similar across the various products, and all tended to follow the WHO guidance [16] on the assessment of tumourigenicity of mammalian cells (i.e. 1x106 cells, subcutaneous administration, 3 to6 months in duration), except that intravenous administration was often used as well to better reflect the clinical RoA. Interestingly, this WHO guidance was developed for characterisation of a master or a working cell bank (MCB, WCB) [16], and not to address the tumourigenic potential of human cell-based products. The suitability of this study design for evaluation of the tumorigenic risk of cell-based products has not been established. Recently, the added value of in vivo tumourigenicity studies has been questioned [21]. With respect to this, it is interesting that only for a minority of products the in vivo study was considered to have some relevance by the regulators or experts. Despite the limited relevance, the non-clinical data package on the tumourigenic potential was considered sufficiently informative for most of the products. This implies that while limited value was given to the in vivostudies, the value of in vitro studies for the assessment of tumourigenicity is recognized. Notably, an in vitro -only evaluation of tumourigenic risk was accepted for two types of products, i.e. for autologous cells and for genetically modified products. For the genetically modified products in vitro evaluation of insertion site analyses (ISA) seem to be sufficient for addressing the tumourigenic risk. Possibly, the need for in vivo animal studies for the safety assessment of cell-based products may be further reduced and more sensitive and standardised in vitro assays may be developed to characterise the risk for tumourigenicity [18,21,30].
To conclude, our analysis has given insight into the non-clinical development program for cell-based products, specifically on the number and types of studies performed for the various types of products. It appears not possible to define a common route to-be-followed for the non-clinical development program of cell-based therapies. While it is clear that during development studies on the pharmacology, biodistribution, general safety and tumourigenicity should be considered, it is not evident that for all these aspects animal studies need to be performed or how they should be designed. Not only because the variety of products prohibits a-one-size–fits-all approach for non-clinical development, but also because insights are changing and clinical experience is growing, which can affect the need for animal studies for future products. Moreover, technical possibilities to study certain aspects may change in time as well. Therefore, the recommendation remains to tailor the non-clinical development program for cell-based products to the specific need for information, depending on the characteristics of the product and take into consideration available knowledge and relevance of in vivo animal studies. To this end, early dialogue between sponsors and regulators will remain utterly valuable.
Acknowledgement section