Production capacityand work flow of MSCs for COVID-19: Using 2D culture system
With Remestemcel-L being so close to FDA approval for ARDS. I thought I will spend some time to work out the biggest bottle-neck of our therapy: manufacturing. Today, I have a go at determining how many doses of cells can we manufacture in a week. The number was quite surprising, toward the upside!
1) Firstly,I need to know how many cells (on average) are needed per patient. An average male in US is ~200 lbs, while female is ~170 lbs; together with the probability of intubation of 60% (M) and 40% (F), come to an average of 188 lbs or 85kg. We know that patients will be given 2 doses of 2 million cells/kg => 85 kg x 2 doses x 2 mil = 340 million cells (or 0.34 billion cells per patient).
2) FromLonza website, under 2D culture technology: the following 2D systems are available, namely CellSTACK, Hyperstack and Cell Factory. After doing some research, CellSTACK is an older technology; whilst Cell Factory system (form Nunc) has a large cell surface area of 25,280 cm sqr, but it is very bulky and will require a much larger space (thus cost) for production. This lead to me assume that if possible (big IF MSC can be cultured in this flask), MSCs will be cultured in a hyperstack system
https://pharma.lonza.com/cell-and-gene-therapy?q=%2Fcell-and-gene-therapy
3)The number of MSCs per hyperstack:
Based on Corning, each hyperstack has 36 layers each having a surface area of 500 cm sqr, totaling 18,000 cm sqr. From info in the video (https://www.youtube.com/watch?v=EfPjal5jKmk), 1 flask could grow up to 2 billion cells. I’ve also double checked the optimal seeding density of MSCs to be 10,000 cell/cm sqr. “https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6431404/pdf/SCI2019-7012692.pdf”. From experience 10,000 cells/cm sqr is very sparse and typically the cells can grow 10-fold before harvesting. At the density of 10,000 cells x 10-fold x 18,000 cm sqr of the hyperstack = 1.8 billion cells, which is in line with what the company provide.
4) Thus wehave worked out 1 hyperstack flask is enough for 5 patients (round down from 5.3 to account for cell loss). Corning mentioned that you could link 100 of these hyperstacks together to generation 200 billion cells per batch (video from above)
5)The more realistic work flow and some extradetail:
Although one could theoretically link 100 hyperstacks together, it will not be physically possible to process that many flasks all at once, which could lead to massive reduction in yield. A wiser and likely solution is to link 6 hyperstacks together and process them using an “Automated Manipulator Platform” (https://www.youtube.com/watch?v=1BwjUxO2FhI). By processing 6 flask at a time, we should be able to ensure the cells are healthy when seeding and harvesting.
6)The simulated workflow per “Automated ManipulatorPlatform (AMP)
Assume that we have prepared the cells and ready to put them into the 6-linked up flasks on an AMP, let’s assume this process takes 1 hour to complete, and we work 2 shifts of 14 hours per day total: 6 flask x 14hr = 84 flask per day. However, since MSCs are relatively slow growing cells that have a doubling time of 40 hours (see image below from Thermofisher), requiring a total of 3.25 x doubling time or 130 hours (5.42 days) to grow 10 times. Assume we need to change media (feed the cells on day 3, harvest day 5.4), we will again require to spend the whole day of AMP usage for changing media. Based on my prediction the final work-flow is:
Day 1: Put cells in culture (lot 1)
Day 2: Put cells in culture (lot 2)
Day 3: Change media for lot 1
Day 4: Change media for lot 2
Day 5: Harvest lot 1
Day 6: Harvest lot 2
Day 7: Prep for next lots
![https://hotcopper.com.au/data/attachments/2132/2132892-51c28c041b51f05832f62d9b9a0f9b4a.jpg](https://hotcopper.com.au/data/attachments/2132/2132892-51c28c041b51f05832f62d9b9a0f9b4a.jpg)
In conclusion:
Based on my conservative assumptions, we could produce at least 2 x 84 = 168 hyperstack flasks per week per AMP. This is enough for 168 x 5 = 840 patients. Based on Bell Potter analysis of peak sales of $520 mil at an average price tag of $67,000, equating to 7761 patients. We could easily fulfil this number of doses using onlyone AMP in 10 weeks!!!
From this exercise, Isuspect that manufacturing for the most critical patients at current doses forCovid-19 is not a big problem. Of course, I have not factored in harvesting and expanding the cells from donor, which will take time to expand enough before seeding into the hyperstack, nor have I factor in the harvesting time, washing cells, centrifuging time, storing in freezing media, aliquoting cells into tubes, labeling, storage and transport.
I don’t have any experience working in a manufacturing plant nor do I work for Lonza. My calculations are all based on the literature and my personal experience as a biologist. These information is for personal use only, not investment advice.