Abstract
To date, cell and gene therapy (CGT) drugs have predominantly been developed to treat relatively small patient populations and utilized as late-line treatments. As such, the field has yet to experience the demand for manufacturing capacity akin to that of “blockbuster” drugs. In this analysis, we utilized our proprietary quantitative modeling platform, Pegasi, to explore whether, in its current state, the field could support commercial supply of a CGT drug at a blockbuster scale. By utilizing the Pegasi platform to model two “prototype” CGT drugs, we created manufacturing capacity projections for each drug at a blockbuster scale, as well as modeled the impact of presumptive advanced manufacturing technology introduction on the capacity projections. The analysis indicates that, in the current state, a successful commercial launch of a CGT drug would be challenged by the lack of available unfragmented manufacturing capacity, as well as the availability of an appropriately skilled workforce. Introducing advanced manufacturing technologies can meaningfully alleviate these challenges; however, developers would still need to make significant investments to close these capacity gaps, putting further pressure on already borderline prohibitive COGs for these therapies. Thus, a new question emerges: does CGT need to fundamentally rethink its current manufacturing and distribution model?
Introduction
It has been widely discussed that, despite examples of “billion dollar” CGT products like Zolgensma and Yescarta, significant growth of the CGT industry will hinge on delivering successful commercial products. While challenges rolling out commercial products have been attributed to clinical site readiness, reimbursement coverage, and accessibility, the fact that CGT drugs continue to predominantly target relatively small patient populations (e.g. rare diseases, late-line therapies) has not yet challenged the field with an important question: is the CGT market ready to support a roll-out of drugs targeting large patient populations, akin to traditional “blockbuster” drugs? For example, for the most recent CAR-T product approvals (Carvykti, Breyanzi), the annual batch production three years after commercial approval was approximately 10x greater than the annual batch production during the pivotal trial. While these products continue to ramp up their commercial production, and sales are forecast to increase 7-fold by 2030, this still represents a total annual patient population of less than 10,000 per product. In contrast, the top 5 selling biologics (neglecting vaccines) average over 300,000 annual patients each, all requiring multiple doses per year, representing a logarithmic increase in demand over the current state.
This paper explores scenarios in which CGT products are approved for a patient population similar in size to that of a first-line blockbuster pharmaceutical drug. To help guide manufacturing decisions, DHC developed a proprietary quantitative modeling platform, Pegasi. Using this platform, the feasibility of manufacturing commercial CGT products at CDMOs at a level that would be comparable with blockbuster biologics was evaluated. A facility mapping exercise was performed to estimate manufacturing capacity requirements as the production targets grew from current CGT clinical and commercial levels and beyond to typical annual targets achieved by other biologics. The analysis considered two “prototype” products: an autologous CAR-T and an off-the-shelf AAV with systemic administration. For each of these products, a representative process, analytical testing, equipment, and labor required to complete production were established. For the AAV product, the scale-up of the process from a 200 L to a 2,000 L bioreactor was also considered.
Case Study: Off-the-shelf AAV
AAV gene therapy products are typically manufactured with a platformed process. This product class also has the advantage that it can be produced independent of patient enrollment and stockpiled for later use. These manufacturing and distribution processes are, in essence, a “best-case scenario” for CGT products. Our model determined that production of the AAV product at blockbuster levels required significant facility capacity, approximately 1M sq ft (Figure 1). While this is undoubtedly a large facility space, it isn’t out of the range for current pharma companies. However, most pharma companies are making dozens of commercial products, and this space would need to be dedicated to a single product. Additionally, the workforce allocated to this facility would be approximately 4,600 full-time employees (Figure 2). Despite recent layoffs, the CGT field as a whole still faces a shortage of skilled staff, placing even more pressure on that bottleneck.
Figure 1. Facility capacity needs for AAV production targets: Total facility space considers cGMP spaces as well as supporting analytics, lab, office, and on-site warehouse and storage. The orange line represents the annual production capacity in a median CDMO facility of 55,000 sq ft.
Figure 2. Related workforce needs for AAV production targets: Total workforce needs include all roles and levels required to release product, which includes manufacturing, quality control, quality analytics, environmental monitoring, supply chain, facilities, operations, MSAT, and supportive admin teams. The orange line represents the annual production capacity in a median CDMO facility of 55,000 sq ft.
Case Study: Autologous CAR-T
Manufacturing autologous CAR-T products is complex and relatively non-industrialized compared to other CGT and pharmaceutical products. As an autologous product, scale-up is not possible, and production will always be 1-to-1 with the patient. Autologous products also require complex supply chains that include dependency on the patient and clinical timelines. This product represents a “worst-case scenario” from an operational and logistics perspective. Even the relatively low patient populations treated by current commercial products require significant facility space in the 100,000s of sq ft. As low patient populations grow to typical pharma and blockbuster pharma levels, the facility space needed becomes immense, with our model predicting that over 8M sq ft of facility space would be required to produce at the same volumes as a blockbuster biologic (Figure 3). In line with facility needs are exorbitant labor needs, with our model predicting a need for nearly 36,000 full-time employees. This essentially would be the entire workforce of Astellas or Boehringer Ingelheim committed to the production of a single product.
Figure 3. Facility capacity needs for CAR-T production targets: Total facility space considers cGMP spaces as well as supporting analytics, lab, office, and on-site warehouse and storage. The orange line represents the annual production capacity in a median CDMO facility of 55,000 sq ft.
Figure 4. Related workforce needs for autologous CAR T: Total workforce needs include all roles and levels required to release product, which includes manufacturing, quality control, quality analytics, environmental monitoring, supply chain, facilities, operations, MSAT, and supportive admin teams. The orange line represents the annual production capacity in a median CDMO facility of 55,000 sq ft.
Discussion
This analysis highlights a big concern – how can these facility and labor resource needs be met? To answer this question, we referred to our proprietary CDMO database of over 250 sites capable of producing CGT products. Utilizing that database, we looked at production capabilities separately for cell and gene products. We compared this to the needs of a single blockbuster product described in the above case studies (Figure 5). This very clearly points out that, especially for an autologous CAR-T, the facility capacity cannot come exclusively from CDMOs as even the combined 104 facilities known within our database to have cell or gene-modified cell production capacity still fall over 2M sq ft short to manufacture a single blockbuster autologous CAR-T product
For the better case of the AAV product, at first glance, it appears that there is sufficient manufacturing capacity. However, considering the median facility size for the 118 vector CDMO within our database is 55,000 sq ft, this would require the capacity to be heavily fragmented across over a dozen sites/providers. This strategy could be more challenging to utilize at scale in a real-life scenario, but companies engaging in decentralized manufacturing strategies have shown that the difficulties of managing multiple sites and site-to-site comparability can be overcome, and will become more approachable as the industry matures. .
Figure 5. Total CGT CDMO capacity compared to blockbuster need: Total CGT CDMO capacity is insufficient to meet blockbuster need (CTx or GMCTx) or would be required to be heavily fragmented across over a dozen sites or providers.
In contrast, the therapeutic developers could build out manufacturing capacity internally. This has been the more common route taken by CGT companies achieving commercial approval, since the field is young and commercial production is still a new demand. This build vs buy decision is one we assist clients with regularly, and by leaning on our Pegasi model again, we can identify hurdles to this approach, namely the large capital investment required. The total installed cost (includes construction, equipment, and indirect costs) for a 1M sq ft facility for the blockbuster AAV product is estimated to be approximately $1B while the cost for the more than 8M sq ft facility for the blockbuster CAR-T could be more than $9B.
In reality, the case of a blockbuster drug would likely require a combination of production strategies. Multiple models to achieve manufacturing capacity are already being investigated, the most common include:
Recommendations
Given the large amount of manufacturing capacity predicted to be required by this analysis, regardless of the decision to build, buy, or another strategy, there would be additional benefits to decreasing the facility, labor, and financial resources needed for manufacturing in the first place. This is especially relevant for autologous products, and the movement toward allogeneic products will only relieve a portion of the burden of a resource-intensive manufacturing process. Below are examples we modeled using the CAR-T prototype to quantify the effect of manufacturing technology evolution and streamlining processes.
Introduction of technology to reduce facility capacity
This will primarily be process automation utilizing end-to-end manufacturing technologies. This automation would need to reduce both the space needed for human movement and the equipment needed for the manufacturing process. Another method could be process intensification to achieve higher throughput within a given footprint. This could result from increasing batch yields typical of scale-up and optimization or a paradigm shift such as an allogeneic instead of autologous product. Higher throughputs can also be achieved by reducing the process duration, a common approach for many CAR-T developers. At the extreme, if the manufacturing capacity required could be reduced to a single, small piece of equipment, this would greatly affect the feasibility of point-of-care or other decentralized manufacturing models.
Companies developing end-to-end manufacturing technologies claim to achieve 30-90% reduction in facility capacity. These technologies range in size from fully automated rooms to smaller table-top units. Targeting the median size, we modeled a theoretical system, similar in size to an incubator, that could similarly be stacked 2-high in a cleanroom. A typical sized cleanroom, approximately 600 sq ft, would contain multiples of these units with additional benchtop space and small equipment. Our model predicts a total facility reduction of approximately 40% and a reduction in production space (Grade B, C, and D) by approximately 80%, as seen in Figure 6.
Figure 6. Impact of manufacturing automation introduction on capacity requirements: Utilizing end-to-end manufacturing technologies to implement process automation provides an opportunity to significantly reduce facility capacity requirements.
Introduction of manufacturing technology to reduce workforce
Unsurprisingly, methods to reduce facility cost and capacity needs also reduce workforce requirements. Process automation moves workforce needs away from operator-intensive, manual manipulations toward a “set it and forget it” operational strategy where operators would only be required to initiate and terminate the batch production. It’s worth noting that as production becomes more automated, this often pushes the workforce bottleneck onto the QC team and highlights the need for analytics automation. Production and analytics technology often enable digitization of workflows if the technologies are selected to be compatible or resources are allocated during development to create that ability. A truly advantageous automation technology would benefit all teams – manufacturing, QC, and QA.
Technology developers have claimed to reduce labor needs by 50-90%. In our model, we assumed that operators would be needed to initiate and terminate batch production, but all operations on other process days would be eliminated. This reduced manufacturing tech time from 13 to 35 hrs per batch and reduced the total manufacturing team by over 87% and labor at the manufacturing site by approximately 58%, as depicted in Figure 7.
Figure 7. Impact of manufacturing automation introduction on workforce requirements: Process automation moves reduce workforce requirements due to fewer manual manipulations required by the operator in place of operational strategies where the operators are required to initiate and terminate the batch production.
Ultimately, a combination of the technology and manufacturing advances noted above will be required for CGT products to compete with traditional biologics. Developing a strategy that faces these challenges head-on requires forethought and planning that takes into account the intricacies and nuance of the manufacturing and operational methods.
