Non-clinical package
In total 218 ATMP advice reports from 2013 until 2018 were collected from the EMA database, containing information on 153 different products. Of these, 59 were gene therapy medicinal products (GTMPs) not consisting of cells, which were excluded from the study. One product was excluded from the database, because it contained xenogeneic cells and seven products were excluded, because information on the non-clinical program was lacking. The remaining 86 products were categorized based on cell origin and on genetic modification status (Figure 1). The majority of cell therapies in development for which advice was requested during the 2013-2018 period, were not genetically modified (n= 60, 70%). Of these,31 were of autologous origin and 29 were of allogenic origin. The majority of the genetically modified cells were autologous of origin (n=20, 23%) and only six (7%) products were genetically modified, allogenic cells.
In total 234 in vivo studies were performed for 80 of the 86 products, and for some products additional animal studies were envisaged. The six products for which in vivo animal studies were not conducted, planned or requested did not contain genetically modified cells (data not shown). Pharmacology, which mainly accounts for proof-of-concept (PoC), was studied for most of the products, followed by safety (Figure 2). The number of products with biodistribution and tumourigenicity animal studies was slightly lower. For these types of studies information on biodistribution (8 products) and tumourigenicity (14 products) was not always present in the non-clinical data package provided by the sponsor. It is not known whether sponsors did not regard studies addressing biodistribution and tumourigenicity of value for their product or whether the product was not yet in the stage of development that the issue of biodistribution and/or tumourigenicity study would be relevant to address.
In guidelines for ATMPs it is recommended to combine the various endpoints addressing e.g. pharmacology and safety in one study if possible and sensible, resulting into a so-called combination study [7]. Across the different types of in vivo studies, pharmacology was most often evaluated in dedicated studies, whereas safety was often studied in combination studies (Figure 3).
Biodistribution
For at least half of the products the sponsor regarded an in vivoanimal study of value for the evaluation of the biodistribution of their product, as for 49 of 86 products (57%) a biodistribution study was either conducted (n=40; 47%) or planned (n=9; 10%) (Table 1). For 25 products, the non-clinical data package contained two or more in vivo biodistribution studies. For the products (n=37; 43%) for which a biodistribution study was not conducted nor planned, this was agreed upon by the regulators either immediately (n=21) or when supported by further justification (n=3). For five products (6%) the regulators requested a biodistribution study in animals. For eight products (9%), a discussion on the biodistribution was lacking in the sponsor’s documentation and was not discussed by the regulators in the advice report. Thus, for 24 of 86 products (28%), both sponsors and regulators considered there was no need for in vivo studies on biodistribution. There were no large differences in the presence or absence of in vivo biodistribution studies between the product categories (Table 1). Even though for most products biodistribution was studied in a dedicated study (30 of 49 products), sponsors indicated that the results were of value for both understanding the PoC of their product and its safety profile. For several products the value of the biodistribution study for PoC-only or for safety-only was recognized (respectively n=4 and n=11).
For nine products biodistribution was studied in healthy and for twelve products biodistribution was studied in diseased animals. For nine of these it was explicitly noted that the model was homologous. For three products no information was present on the model and for the remaining 25 products immunocompromised animals, reflecting heterologous models, were used. For 42 products information on the method to evaluate biodistribution was provided. The most commonly used techniques were IHC, quantitative-PCR (qPCR) or reverse-transcriptase PCR (RT-PCR), Fluorescence In Situ Hybridization (FISH), and microscopy. For all these techniques tissue samples need to be harvested to allow for detection of the cells. The more sophisticated techniques for studying cell-distribution in vivo (such as Positron emission tomography (PET), Single photon emission computed tomography (SPECT), Magnetic resonance imaging (MRI) and Bioluminescence imaging (BLI)) were only used to study biodistribution for three products.
The vast majority of biodistribution studies provided information on the presence of the administered cells in/migration to both target and non-target tissues (40/49; 82% of the products with biodistribution studies, Figure 4). Only for four (autologous cell) products, sponsors considered the option of assessing target tissue only (8% of the studies). For five products it was not mentioned which tissues were assessed (10%). Information on the relative distribution of the administered cells was available for 42 (86%) of the 49 studies, suggesting that the method to determine biodistribution of the therapeutic cells was also (semi-)quantitative.
Information on persistence of the administered cells was available for 44 products (51%). For another 10 products (12%) sponsors planned to evaluate in vivo persistence in the future, whereas for 13 products (15%) sponsors did not aim to address it (Figure 5). For the remaining 19 products there was nothing mentioned regarding persistence in the submitted information.
Tumourigenicity
Remarkably, while FDA guidance [11] specifically stipulates that for tumourigenicity studies the intended clinical product should be used (and not analogous animal cells), tumourigenicity was studied in a homologous model for eight products of which three were studied in healthy and five in diseased animals. For one product it was not mentioned and for remaining 37 products immunocompromised animals, reflecting heterologous models, were used. The chosen route of administration (RoA) was most often intravenous or subcutaneous and mainly a dose of 1x106 cells was applied (Table 2). For two products, various doses were used and for ten products the administered dose in the study was not mentioned (most often for planned studies). While there was some variability in the duration of thein vivo tumourigenicity studies, a study duration between three and six months was noted for the majority of the products (n=28 of 46).
For most products the tumourigenic potential was addressed by experimental data (in vitro and/or in vivo studies), only for 10 products (11%) sponsors did not investigate the tumourigenic risk with any experimental data. Of note, for 14 (16%) products there was no information on tumourigenicity studies in the documentation. For nine products (10%), the risk for tumourigenicity was investigated byin vivo studies only, for 18 products (20%) the tumourigenic risk was only evaluated by in vitro assays, and for 37 products (43%) both in vivo and in vitro studies were used to address this potential risk (Figure 6). Thus for 46 of the 86 products in the database tumourigenicity was studied and/or planned to be studiedin vivo (Figure 2). For 18 of the 46 products with in vivotumourigenicity studies, more than two in vivo studies tumourigenicity studies were part of the development package. This often concerned the combination of a dose range finding study and a pivotal study, or the assessment of tumourigenic risk upon various routes of administration.
For 16 of the 40 products for which no in vivo study was performed or planned, it was explicitly agreed that an in vivostudy was not necessary, while for six products regulators indicated that the absence of in vivo studies could be acceptable provided that the company justified the absence of such a study. This was apparently most often the case for the genetically modified products (Table 3). For four products an in vivo animal study was explicitly requested. Products for which tumourigenicity was studied with only in vitro studies most often contained genetically modified cells (Figure 6). For products containing allogeneic genetically modified products, tumourigenicity was always discussed
The in vitro tumourigenicity studies in the 55 (64%) products most often consisted of karyotyping (n=28) and/or colony formation/proliferation assays (n=22). For 22 products, multiple types of in vitro studies were performed (Table 4).