International Stem Cell Banking Initiative ISCBI PSC 2019

International Stem Cell Banking Initiative (ISCBI) PSC 2019, June 30 th 2019, Los Angeles, CA. Closing Panel: Organizers’ Summary Chairs: Glyn Stacey (ISCBI) and Stephen Lin (CIRM) Participants: Ngaire Elwood (ISCT), Lucilia Mouriès (HESI CT-TRACS), Yoji Sato (NIHS; HESI CT-TRACS), Stephen Sullivan (GAi. T)

Closing Panel Glyn Stacey Director Ngaire Elwood Director, BMDI Cord Blood Bank Lucilia Mouriès Scientific Program Manager ISCB I Stephen Lin Stephen Sullivan Yoji Sato Senior Science Officer International Liaison Officer Head of Division, Cell. Based Therapeutic Products PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA.

Spectrum of activities to make a PSC therapy product Primary Donor Material i. PSC generation Final Differentiated Product Autologous • No donor eligibility determination • Limited material testing Allogeneic • Donor Recruitment and Appropriate Donor Selection • Donor consent determination • Screening and testing required within 7 days of harvest • Traceability of reagents & materials • Regulatory approval • Master Cell Bank • Testing • Reprogramming/Gene Modification/Editing PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. • • Materials Manufacturing process Regulatory approval Product stability Safety Testing Efficacy Testing Storage Final formulation 3

Panel 2: Pluripotent Stem Cells as Manufacturing Cell Substrates - Challenges Encountered and Forecast Pre-recorded interview for this panel: PSC/i. PSC therapeutic manufacturers should for specific applications to control variation. consider the clinical usability of their product from the beginning of the translation Basic researchers do not intuitively pick the best CQA for scale up on their own, process. Besides safety, efficacy, and cost, usability is be a key parameter determining overall clinical impact. If the administration of therapy will be too every process change made must consider the potential change to the final product. Changes in marker expression or functional assays chosen should be indicative of a complex, requiring too much specialized knowledge, it will not be feasible to administer at many locations: consequently therapy will not have a significant quality change to the final product. clinical impact. – Dr Bill Shingleton (HESI CT-TRACS). Even if a starting material meet the same quality attribute definitions, the product Understanding and reducing process variation: need to understand genetic drift of may turn out to be very different due to culture/scaling systems. material. Define CQAs at all stages of development form beginning to end Should we be focused more on mechanisms of action before selecting Critical Demonstrating comparability: comparability tests change depending on chronic vs Quality Attributes? acute condition Challenges of i. PSC therapeutic manufacturing: comparability, cost of CTP, length of Biggest challenge for pluripotent stem cell derived therapies: understand how they induced differentiation processes. work (not just paracrine effect) Transition from process centric cell therapies move to product centric cell therapies is desired. Controlling and characterizing the starting material is very important. Quality control for the starting material is currently a challenge. More biomarkers are required. Drawing experience from h. ESC induction into cardiomyocytes, there is significant line variation. We need to understand all forms of variation. With biological variation for example, we need to understand better understanding of Inter-clone variation vs inter-line variation vs inter-donor variation. Can we design a manufacturing process that is flexible enough to allow for the biological variation exhibited between donors, and how much can we broaden the definition the process parameters before it becomes a significantly different process that has to be qualified or validated separately? Build robust process so it can accommodate many lines, or else pick specific clones PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 4

Panel 3: International Standardization The difference in incentives to share between companies and public bodies was highlighted – public bodies developing of clinical trials would be more open to data published at the beginning. The difference between having information in the public domain and having the information is a usable/searchable format was Pre-recorded interviews for this panel: (i) Standardization of i. PSC lines as Starting Material for Therapeutic Development – Prof Marc Turner (GAi. T, SNBTS) and (ii) Efficiencies of PSC/i. PSC translation effort are assisted highlighted. FDA clinical trials have extensive information published but it is very tedious to go through such information. For example, with gene therapies the clinical information is structured around go/no go decisions by the different professional silos being aware of the others’ activities and challenges. The properties a PSC/i. PSC cell line must have to gain regulatory approval as a starting material for manufacture, for example, related to allowing clinical trials to proceed in specific jurisdictions so there is considerable variability in how the data is presented. may not always be discernible to academic colleagues and these are helpful for all focused on clinical translation to be aware of – Prof Marc Turner (GAi. T, SNBTS) Sharing enough information to tool developers will be needed to allow them to develop better characterisation assays that are currently needed. It can be difficult to share technical information while still Data sharing protecting business sensitive information. Panel 3 Pluripotent stem cells as manufacturing cell substrates – challenges encountered and forecast Clinical: h. PSCreg database now developing, welcome input Challenges with EU/GDPR regulations new developments in other countries. Consents: ISSCR standards, still issues with institutional variation The route forward appears to be to continue highlight the benefits of data sharing especially at a time where efficacy data for PSC trials is still being sought. Ethics and General Data Protection Regulations (GDPR) should be considered carefully – special provisions will be needed Best practice for consent is still evolving. Validating the provenance and constriction for downstream use of lines is important. ISSCR has produced a consent template which has the basic elements for consent but is really designed to incentivize lawyers to better combat medical tourism. h. PSCReg clinical trial database Linking data from different systems is a challenge. May need new kinds of reference materials. Standards will information validation would allow better description from good and bad actors. GDPR is not just a concern for improve regulatory acceptance. those operating in the EU. Samples, lines, and data merely passing through EU would be affected and special provisions like (private by design) will be required. The increased sharing of process and testing information in order to drive PSC manufacturing standardization and innovation. Commercial interests fear losing competitive advantage if they share manufacturing and Rapid increase in novel technologies for analytics brings further complications testing data but are there other data that can be shared to assist progress in the field. For example, clinical trial data is normally placed in the public domain at sites such as clinicaltrials. gov, but presently it can be difficult to Biological variation is considerable. Process variation is increasing due to the wide range of new analytics find clinical trial-based data for PSCs. becoming available to deal with the complexity of the cell biology. Standardization with high content analytics: Understand biology in early development. The h. PSCReg generated database for PSC clinical trials was introduced. Sharing data is not the same as sharing High content analytics hopefully can assist with identification of CQAs practice, and if efforts to develop PSC therapeutics are to standardized, it is important to understand what processes are actually being used. Image analysis of i. PSC cultures are very critical. Companies developing software for removal of operator Prior to generating the clinical PSC database, Andreas Kurtz identified over 40 clinical trials using embryonic variation. ISO developing for image analysis. stem cells and pluripotent stem cells, and the information deposited in his database is validated, unlike Stakeholders are encouraged to participate in standards development discussions so standards are fit for clinicaltrials. gov. In addition to clinical trial information, it is proposed to make PSC data more accessible; additional information fields be added to capture what cells are being used in the end product and how the purpose end product is characterized, these would include potency, sterility, efficacy, and mode of action (if known). Such data needs to be mapped, in addition to other information fields like genetic data (from karotyping to rt. PCR for PSCs. whole genome sequencing) particularly of the donor and starting material. Genetic data collection and storage in particular is complicated by security and personal data protection issues. Follow up clinical data should also All stakeholders including researchers should participate in standard development so the standards are be mapped. feasible/acceptable/suitable. NIST/SCB/ISO all active in this area. Artificial Intelligence for Data Platforms and Systems PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 5

Panel 4: Defining Quality Parameters as manufacturing intermediates • Potency for PSCs as an intermediate: Look at end product. Cardiomyocyte vs neuron. Test your final product. Break up process and isolate tests • There is an important distinction between information use only vs safetyrelated testing when evaluating pertinent Quality tests From a PSC banking perspective, there is a desire to provide material that has been screened by canonical tests that can provide downstream users more confidence in the potential performance of the material. However, a test like cell marker expression has no apparent bearing on clinical safety; therefore, this attribute is merely captured for information use only and not considered mandatory for regulators. The ability to confirm h. PSCs do not harbor detectable m. RNA, virus, and/or plasmids does, however, provide • relevance to clinical safety. Therefore, qualified tests designed to evaluate potential for residual reprogramming elements would be desirable to regulators. While there is regulatory guidance for more routine tests such as sterility and endotoxin, custom tests specific to the desired product are unlikely to have specific guidelines associated with them. Therefore, manufacturers should not expect the FDA to provide customized guidelines rather they should seek guidance early on from the FDA during the assay development process. The FDA is receptive to customized testing if the rationale for the test remains safety focused, backed by a body of solid data, and has been sufficiently validated. • Cell authentication requires clarity on which reference material is used to confirm appropriate match to resulting h. PSCs. There are many case reports where h. PSCs do not match the intended material. A question remained whether current methods to evaluate identity require development to enable distinction amongst clones from the same starting material. Is it sufficient for GMP facilities to minimize and segregate clones during manufacture or does there remain potential risk for contamination? Participants reinforced this relevance as increasing numbers of gene edited h. PSCs emanating from the same source material increasingly seed the field. A counter point reinforced the need to keep CQAs simple and direct and raised awareness that more information could lead to unnecessary information that may be indecipherable and limit progress into the clinic. The use of h. PSCs manufactured completely in GMP versus those derived in R&D then transitioned to GMP remain a source of debate with reasonable rationale on either side. It was generally agreed that the level of rigor employed by GMP manufacture affords the best option to minimize patient risk. However, the costs for GMP manufacture can be prohibitive making this an intractable option for many and reducing the valuable work necessary to inform future therapies. Consequently, there have been several cases where R&D grade material has been sourced and transitioned (e. g. h. ESCs) into GMP for clinical trials. It is important to keep in mind that variable "grades" of R&D material exist that are overseen by good Quality Management Systems. Therefore, appropriate due diligence and/or auditing would be advised to evaluate the integrity of material received. This topic remains a source of debate and further discussion. Identity was discussed to provide a specific CQA for panelists and audience members alike to contemplate. While it was unanimously agreed that Identity testing is critical, it facilitated a deeper discussion that involved cell authentication and debate on whether further method refinement is necessary to track clonal populations. PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 6

Panel 5: Roadmap for comparability • Donor material: Keep as much as possible to test. Emphasize donor eligibility. • Testing: Take risk-based approach. Try to share information. Learn from failures. Develop robust protocols. PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 7

Panel 6 - tumorigenicity • Diversity of testing methods; no consensus on best practices. • Better characterization of testing methods are critical for the translation of i. PSC-derived products. • Identifying ideal positive control cells for predictivity of clinical safety is a challenge. PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 8

HESI Cell Therapy-TRAcking, Circulation and Safety (CT-TRACS) Committee lmouries@hesiglobal. org www. hesiglobal. org 9

Panel 7 – Automated and closed manufacture Enabling technology development • Key aspects of automation: Expansion Want mature process before addressing this • Variable that predicts risk for automation development: Need to separate lines. PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 10

Panel 8 – Financial and logistics challenges • Logistics costs/challenges: TPP, quality by design They are high, but good therapeutic will justify costs. Difficult to hire and keep manufacturing experts • Not thinking about IP strategies PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 11

Panel 9 – Immune rejection/engraftment failure Pre-recorded interviews for this panel: Jeanne Loring (Aspen Tx) discusses autologous cell there is a problem (i. e. tumor formation) you know longer have the default system for therapy development and how while it might not be as scalable as allogenic approaches, it dealing with such undesirable cells. This would necessitate the introduction of an inducible will probably be the best approach to identify proof of concept, David Turner (SNBTS) cell suicide system which comes with further complications. Taking away immune system discusses the current state of transplantation immunology and how even the clinical capacity to respond for safety purposes may mean you are actually not making your immunologists do not know everything, Marc Peschanski (INSERM) discusses their process any safer overall. upcoming i. PSC clinical trial for AMD and the considerable amount of work required to get regulatory approval. Universal donor approach is not a magic bullet yet and may not ever be. Immune systems not fully characterized. In many cases we do not have good models to anticipate immune • Immune rejection remains a formidable barrier to stem cell transplantation. Stem cell response. Different models will be useful in different cases. biology has assisted the interrogation of immune rejection after tissue transplantation but immunology can assist with making stem cell derivatives less likely to be rejected Also CAS-9 introduction itself can cause immune response, but this would be a problem is after transplantation. CAS-6 persists in your stem cells (genetic integration). Either transient expression or m. RNA introduction should be all that is required for engineering proposes so the other challenges • Immunosuppressive agents, especially when chronically administered, have serious will probably be more clinically significant. side effects. Impeded immune function can lead to a greater cancer risk (through lowered onco-vigilance) and also a greater risk of opportunistic infection (opportunistic • Engraftment encapsulation, choosing immune-privileged sites (brain, eye) and microbes are not normally a problem but can be a significant clinical complication immune matching (haplobanking) are other approaches. Karolinska Institute is during reduced immune function). Strategies to reduce the clinical reliance on such adopting a combinatorial strategy: haplobanking approach/ cell engineering/ non-specific agents should be welcomed. transplantation into an immune-privileged site. Many publications over the last year dealing with genetic modifying cells so that they are • less likely to trigger an immune response. The so called ‘universal donor’ strategy submits that through such editing, an engineered cell could potential be introduced into any patient without immune rejection. At a shallow level of interrogation at least, such an approach • seems elegant, quick and the most prudent. However, once considered in detail, it may not be the magic bullet it appears to be. The more novel technology used, the more safety testing and paperwork will be required to gain regulatory approval. Immune system not fully characterized and our capacity to absolutely predict immunomodulatory element is not perfect. In many cases we do not have good models to anticipate immune response. Different models will be useful in different cases. For example, Universal Cells LLC introduced HLA-E component to stop natural killer (NK) response but this does not stop all NK cells. Chad Cowan sticking on HLA-G expressed instead. Different NK cells have different inhibitory antigens. Might need combination of these approaches and also there is immune rejection not mediated through NK cells. No magic bullet. Each approach has advantages/disadvantages. Immune response dependent on many variables (Cell/Exposure/Response variables). Immune system not fully characterized. Development of humanized model for predicting Approaches may not be mutually exclusive and may be used in a immune response hindered by fetal tissue ban in USA. Off-target modifications, unforeseen combinatorial therapy. consequences of on-target modifications (e. g. capacities to differentiate and/or scale – will these be affected) – if you are engineering the cell, it should be checked that deleterious to the process of therapeutic development itself. Additionally, engineering a cell that cannot be detected by the immune system, means if PSC Manufacturing Expert Panels 2019 - June 30 th 2019, Los Angeles, CA. 12