Aiming at enabling safe, effective and reproducible therapeutic use of donor-derived dermal ABCB5+ MSCs, we developed and validated a GMP-compliant ex-vivo expansion and manufacturing process to generate an ATMP derived from skin tissue donations that guarantees reliable and reproducible cell quality and functionality while being deliverable as an instantly available off-the-shelf product.
From experience with MSCs derived from bone marrow or adipose tissue it is known that age and morbidities of the donor can impair the number, viability, proliferation capacity, maximum life span and functional properties such as differentiation potential and proangiogenic factor secretion [16–24]. For dermal ABCB5+ MSCs, an age-dependent decrease in their number in situ from 3.2% of total dermal cells in the skin of young individuals to 1.6% of total dermal cells in the skin of individuals > 70 years has been observed [12]. Together, these observations prompted us to limit donor age to ≤ 50 years and to define strict donor health requirements. Nevertheless, even MSCs derived exclusively from young and healthy donors have been described to exhibit donor-dependent variations in molecular status, growth behavior and functional characteristics [25, 26], which, ultimately, could result in donor-related batch-to-batch variability. Other factors that could potentially contribute to unwanted product inhomogeneity may be related to variations in the expansion and manufacturing process such as passage number or lab assistant experience. In this context of a situation where the “process is the product” [1, 27], it is crucial to closely monitor the entire process at each step from the starting material until the final drug product to detect and eliminate any factors that would impair the homogeneity of the cell therapy product.
During cultivation, potential donor and/or passage number-related variations in the proliferative capacity of the cultured cells were monitored by assessing the time needed from passaging to reach the target confluency for each passage. In all cases, the specified upper time limits during both expansion (passages 1–6) and subcultivation (up to passage 11, depending on the number of isolation cycles performed) were met (Fig. 2). Low intra-donor standard deviations and generally narrow inter-donor ranges indicate high homogeneity of the proliferative behavior of the cells within and between the different donors, respectively. This is supported by previous cell division data collected during the process validation phase, which revealed a constant mean division rate of 0.35 divisions per day across 15 donors and even up to 16 passages (Kerstan et al., submitted). It should be noted that the number of isolation cycles (which accounted for up to 11 passages at maximum) for each donor depended on the company’s production plan and were, in most cases, not restrained by cell culture-related phenomena such as changes in cell morphology or growing behavior.
Low variation within donors and a high degree of comparability between donors was also seen for the cell cycle distribution of ABCB5+ MSCs (Fig. 3). The phase distribution pattern (82 ± 6%, 3 ± 2%, and 10 ± 4% for G1, S, and G2/M phase, respectively) resembled extensive data collected during process validation (503 analyses from 260 donors: 72.4 ± 8.5%, 4 ± 2.9%, and 15 ± 8.5% for G1, S, and G2/M phase, respectively) (Kerstan et al., submitted) and did not change during passaging until passage 11 (Fig. 3c). Taken together with previous RNA sequence analysis data revealing that the expression of various stemness and mesenchymal marker genes was not altered in ABCB5+ MSCs expanded to high passage numbers (> 10) as compared to low passage numbers (2–3) [6] and microarray data (GEO accession No. GSE145589) showing that the gene expression profile of ABCB5+ MSCs did not significantly change over five delta passages up to passage 16 (Kerstan et al., submitted), these results support that proliferation capacity and MSC properties of ABCB5+ MSCs are maintained during ex vivo expansion up to comparatively late passages.
The isolation process of ABCB5+ MSCs from the expanded mixed cell culture was evaluated by assessing the homogeneity of the ABCB5+ cell isolates and the quality of the drug substance batches produced thereof. Obviously, the cell isolates varied considerably in their mean live cell count between the different donors, with the highest donor mean (48.6 × 106 ± 16.9 × 106 cells per isolate) amounting to more than twofold of the lowest donor mean (19.0 × 106 ± 5.5 × 106 cells per isolate) (Table S2). This donor-related variation was neither dependent on donor age, sex (all donors were female) nor skin tissue localization (abdomen vs. mamma) (Table S2) and may therefore reflect other, likely individual, differences between donors. Variations in the mean live cell count of the cell isolates were also observed between the different lab assistants who performed the ABCB5+ cell isolation. From our experience a certain degree of variation can be expected based on the various processing steps such as harvest, isolation, washing, resuspension and sampling. As it turned out, the degree of variation declined with increasing experience of the lab assistant. In contrast to lab assistant-related variability, the number of passages that a culture had undergone until ABCB5+ cell isolation did not appear to have impact on the live cell count of the isolate, since vitality and viability of the isolated cells did not change with increasing passage numbers (Fig. 5a).
While lab assistant-related variability in the live cell counts between the cell isolates could likely be decreased by enhancement of lab staff training, potential donor-individual influences cannot be eliminated prospectively. Therefore, to limit a potential impact of ABCB5+ cell isolate variability on the homogeneity of the drug substance batches produced thereof, we defined that the coefficient of variation of the cell count of a set of ABCB5+ cell isolates may not be greater than 50% for it to be pooled into one drug substance batch (otherwise only the isolates that did not vary by more than 50% from the mean cell count of the whole set of isolates intended to be pooled were ultimately pooled into a drug substance batch). Applying this criterion, 64 of 66 isolate sets could be pooled as a whole, the remaining two isolate sets at least in part. Thus, in the future this criterion could be defined more strictly in order to further increase process homogeneity. Of note, drug substance batch quality was not negatively affected by this variability. Specifically, only 8 out of 66 drug substance batches were rejected due to specification failure (Table 2).
A key component in the quality assessment of cellular therapy products is the evaluation of potency, in order to guarantee biological functionality that is predictive of clinical effectiveness. Considering that MSCs exert their therapeutic effects through a variety of pathways that are induced upon interaction with the host microenvironment, which has constituted the paradigm of MSCs to function as site-regulated, patient-specific “injury drugstores” [28], evaluation of a single pathway is thought insufficient to predict clinical effectiveness [2]. We therefore developed and implemented three potency assays reflecting the three clinically most relevant biologic modes of action of ABCB5+ MSCs: (i) IL1RA secretion assay after co-culture with M1-polarized macrophages to predict the anti-inflammatory potency in M1 macrophage dominated inflammatory tissue environment, (ii) VEGF secretion assay in hypoxic culture conditions to estimate the pro-angiogenic bioactivity in ischemic tissue environment, and (iii) tube formation assay to demonstrate the endothelial differentiation and blood vessel-forming capacity of the produced cells. Except for 3 batches that failed to clearly form capillary structures in the tube formation assay, all drug substance batches fulfilled the specified criteria for all three assays (Fig. 5b, Table 2). Notably, however, while these assays were validated to reliably detect biological activity and thus qualitatively confirm the potency of ex vivo expanded ABCB5+ MSCs, a quantitative correlation between the assay values and the strength of clinical efficacy remains to be elucidated. Comparisons between the outcomes of the ongoing clinical trials and the assay data results will offer valuable clues on quantitative potency prediction.