I’ve finally read the Huw Hughes review of the CE Delft Techno-Economic Analyses (our summary here) of cultured meat and thought it was interesting commentary on the CE Delft analysis, though less informative on the overall question of cultured meat scaleups than I hoped.
Overall their position on CE Delft’s analysis was similar to ours, except maybe more bluntly worded. They were more critical in some parts and less critical in others.
Things I liked about the Hughes review:
the monoclonal antibodies reference class point was interesting and not something I’ve considered before
I liked how it spelled out the differences in stability between conventional meat and cell slurry; I’ve thought about it before but didn’t seriously consider it, having this presented so clearly was useful
I liked the author diagramming an entire concept area to see what a real factory might look like, instead of only looking at the myocyte cell process
I fully agree with the author about the CE Delft TEA being way too underspecified, as well as the aseptic conditions stuff making food bioreactors a very unlikely reference class (though of course I already knew these points)
I liked the points about regulatory bodies adding costs
I liked the points about more specialized labor presumably being more expensive than CE Delft estimates
Things I liked less:
cell density points was really convoluted compared to our explanation that CE Delft estimates were hitting variously viscosity or literally physical space limits
Hughes assuming that bioreactor design will continue to be a boutique and specialized designs even in an ecosystem with lots of producers and suppliers (Humbird doesn’t do this)
Saying stuff like “reducing costs of recombinants and growth factors will take 4-10 years and several millions of dollars in R&D” even though millions of dollars is like pennies per kg at 100kTA
claiming “prices of $9,000 to $36,000/kg (±30%)” which is a laughably low range of uncertainty
generally, having assumptions that are closer to boutique programs (like existing pharmaceuticals) rather than an ecosystem at large scale
Overall I thought it was interesting and worth reading.
For those interested, here are the paragraphs that added new information on the CE Delft TEA that I hadn’t considered or seen in other TEAs or reviews.
Cell growth
“For bacteria, one may generally subculture in the range of 1:10 to 1:100, depending on the bacterium. For stem cells, which are more fastidious, the highest subculture ratio (i.e. the amount of an old culture one uses to seed or start a new culture) is typically 1:5. This is significantly less than the 1:200 cell ratio that is proposed in the TEA. Experience dictates that if stem cells are sub-cultured at the ratio proposed, they will either differentiate and stop growing, or simply will die. ”
Medium costs
“In an analysis of culture medium costs, Specht (2019) used DMEM/F12 medium as a base, with a cost of $62,400 for a 20,000 L batch (or $3.12/L). However, this appears to be the cost for powdered bulk medium (from, e.g., MP Bio), and does not include the cost for labour, USP water, formulation, filtration and testing prior to culture. Given the fact that up to now stem cells require custom media, the price for base medium alone could rise to $822,000 for the same sized (20,000 L)batch. However, it should be noted that a properly developed medium may rely less on growth factor additives” This is also used by Risner, et al (2020) in their “this is where the miracle happens” Scenario 4.
“insulin costs will likely remain as they are due to current and future manufacturing standards and volumes. ”
Contamination
“The building will require air locks with increasing air handling quality, for both materials and personnel. Typically this comprises a Class D corridor, with Class C rooms except where open phase is carried out, which must be Class A in a Class B environment with Class C minimal changing rooms. The TEA document does not take into account this quality standard, nor does it take into account the additional personnel time”
Stability
“Cell pastes or slurries are notoriously unstable and will deteriorate very quickly,losing their cellular structure. Real meat on the other hand is built on a scaffold of blood vessels, interdigitating cell structure and layers of connective tissues,tendons etc., that helps to maintain structure. Even ground meat will maintain some of this structure at the macro level, and is seldom if ever homogenized to a single cell slurry. Given the large yields of CCP [Cell Cultured Product], a process must be devised to ensure that the slurry maintains its structure.”
Monoclonal antibodies
“In summary, Monoclonal antibody yields have seen a 10- to 20-fold increase in the last 10 to 15 years, with 10s of billions of dollars of investment in R&D across multiple industries. In the TEA example, it is proposed that costs will be reduced, with a resulting >1000-fold decrease in costs. Given the experience with monoclonal antibodies, this may be overly ambitious and does not take into account the fact that every cell bank will be different – it is possible that each one will need to be developed independently.”
Batch times
“In this instance, the TEA authors had proposed a 10-day perfusion culture that would use 800 L of media for each 2,000 L of product. . . . For such a short perfusion time, normally the process would be better suited to a high-density fed-batch process (10-12 d). Perfusion generally is reserved for longer-term cultures (20-35 d or more, Bausch et al., 2019)”
Bioreactors
“Large scale bioreactors (>2,000 L) will remain custom built items for the foreseeable future, and thus will be expensive to build and install.Cost savings initiated through process and genetic engineering to increase yields, cell line development . . . is likely not an option for a multitude of regulatory and social reasons”
Capital costs
“The Capital costs appear to only take into account the myocyte cell manufacturing process. Further, a multiplier of tank capital costs is used to extrapolate to the total capital, rather than drawing a concept design and estimating surface area and cost to within a certain margin of error . . .
Clearly the capital costs are greater than those estimated by the authors in theTEA, with apparently only a small fraction of the equipment and infrastructure accounted for. It is unclear how the ‘Social Investment Criteria’ would work in this situation, as a factory of this size and complexity will cost several hundreds of millions of dollars to build. Due to the complexity of the manufacturing processes, the requirement to remain as an aseptic process, commissioning and validating the plant, even to food grade requirements, could also cost in the millions of dollars, depending on the final product profile. Generally, these costs would be put against the products coming from the factory over a 10-year depreciation period.”
Personnel costs
“Personnel costs are estimated as $100,000/annum/per FTE in the TEA, fullyloaded. This is likely an under-estimate for the operators in the aseptic areas, as staff experienced in the operation, validation and trouble-shooting of complex bioreactor and down stream process processes would be required. This estimate could be increased to $150,000 and even that may be on the low end depending on where the factory is to be located. ”
There are a bunch of other critiques are basically arguing “this is expensive and nobody will fund it”, but that’s just an intuition, not a hard technical stop.
Here’s the layout provided. I’d love to see more of these, like the one WildType provides here
I’ve finally read the Huw Hughes review of the CE Delft Techno-Economic Analyses (our summary here) of cultured meat and thought it was interesting commentary on the CE Delft analysis, though less informative on the overall question of cultured meat scaleups than I hoped.
Overall their position on CE Delft’s analysis was similar to ours, except maybe more bluntly worded. They were more critical in some parts and less critical in others.
Things I liked about the Hughes review:
the monoclonal antibodies reference class point was interesting and not something I’ve considered before
I liked how it spelled out the differences in stability between conventional meat and cell slurry; I’ve thought about it before but didn’t seriously consider it, having this presented so clearly was useful
I liked the author diagramming an entire concept area to see what a real factory might look like, instead of only looking at the myocyte cell process
I fully agree with the author about the CE Delft TEA being way too underspecified, as well as the aseptic conditions stuff making food bioreactors a very unlikely reference class (though of course I already knew these points)
I liked the points about regulatory bodies adding costs
I liked the points about more specialized labor presumably being more expensive than CE Delft estimates
Things I liked less:
cell density points was really convoluted compared to our explanation that CE Delft estimates were hitting variously viscosity or literally physical space limits
Hughes assuming that bioreactor design will continue to be a boutique and specialized designs even in an ecosystem with lots of producers and suppliers (Humbird doesn’t do this)
Saying stuff like “reducing costs of recombinants and growth factors will take 4-10 years and several millions of dollars in R&D” even though millions of dollars is like pennies per kg at 100kTA
claiming “prices of $9,000 to $36,000/kg (±30%)” which is a laughably low range of uncertainty
generally, having assumptions that are closer to boutique programs (like existing pharmaceuticals) rather than an ecosystem at large scale
Overall I thought it was interesting and worth reading.
For those interested, here are the paragraphs that added new information on the CE Delft TEA that I hadn’t considered or seen in other TEAs or reviews.
Cell growth
“For bacteria, one may generally subculture in the range of 1:10 to 1:100, depending on the bacterium. For stem cells, which are more fastidious, the highest subculture ratio (i.e. the amount of an old culture one uses to seed or start a new culture) is typically 1:5. This is significantly less than the 1:200 cell ratio that is proposed in the TEA. Experience dictates that if stem cells are sub-cultured at the ratio proposed, they will either differentiate and stop growing, or simply will die. ”
Medium costs
“In an analysis of culture medium costs, Specht (2019) used DMEM/F12 medium as a base, with a cost of $62,400 for a 20,000 L batch (or $3.12/L). However, this appears to be the cost for powdered bulk medium (from, e.g., MP Bio), and does not include the cost for labour, USP water, formulation, filtration and testing prior to culture. Given the fact that up to now stem cells require custom media, the price for base medium alone could rise to $822,000 for the same sized (20,000 L)batch. However, it should be noted that a properly developed medium may rely less on growth factor additives” This is also used by Risner, et al (2020) in their “this is where the miracle happens” Scenario 4.
“insulin costs will likely remain as they are due to current and future manufacturing standards and volumes. ”
Contamination
“The building will require air locks with increasing air handling quality, for both materials and personnel. Typically this comprises a Class D corridor, with Class C rooms except where open phase is carried out, which must be Class A in a Class B environment with Class C minimal changing rooms. The TEA document does not take into account this quality standard, nor does it take into account the additional personnel time”
Stability
“Cell pastes or slurries are notoriously unstable and will deteriorate very quickly,losing their cellular structure. Real meat on the other hand is built on a scaffold of blood vessels, interdigitating cell structure and layers of connective tissues,tendons etc., that helps to maintain structure. Even ground meat will maintain some of this structure at the macro level, and is seldom if ever homogenized to a single cell slurry. Given the large yields of CCP [Cell Cultured Product], a process must be devised to ensure that the slurry maintains its structure.”
Monoclonal antibodies
“In summary, Monoclonal antibody yields have seen a 10- to 20-fold increase in the last 10 to 15 years, with 10s of billions of dollars of investment in R&D across multiple industries. In the TEA example, it is proposed that costs will be reduced, with a resulting >1000-fold decrease in costs. Given the experience with monoclonal antibodies, this may be overly ambitious and does not take into account the fact that every cell bank will be different – it is possible that each one will need to be developed independently.”
Batch times
“In this instance, the TEA authors had proposed a 10-day perfusion culture that would use 800 L of media for each 2,000 L of product. . . . For such a short perfusion time, normally the process would be better suited to a high-density fed-batch process (10-12 d). Perfusion generally is reserved for longer-term cultures (20-35 d or more, Bausch et al., 2019)”
Bioreactors
“Large scale bioreactors (>2,000 L) will remain custom built items for the foreseeable future, and thus will be expensive to build and install.Cost savings initiated through process and genetic engineering to increase yields, cell line development . . . is likely not an option for a multitude of regulatory and social reasons”
Capital costs
“The Capital costs appear to only take into account the myocyte cell manufacturing process. Further, a multiplier of tank capital costs is used to extrapolate to the total capital, rather than drawing a concept design and estimating surface area and cost to within a certain margin of error . . . Clearly the capital costs are greater than those estimated by the authors in theTEA, with apparently only a small fraction of the equipment and infrastructure accounted for. It is unclear how the ‘Social Investment Criteria’ would work in this situation, as a factory of this size and complexity will cost several hundreds of millions of dollars to build. Due to the complexity of the manufacturing processes, the requirement to remain as an aseptic process, commissioning and validating the plant, even to food grade requirements, could also cost in the millions of dollars, depending on the final product profile. Generally, these costs would be put against the products coming from the factory over a 10-year depreciation period.”
Personnel costs
“Personnel costs are estimated as $100,000/annum/per FTE in the TEA, fullyloaded. This is likely an under-estimate for the operators in the aseptic areas, as staff experienced in the operation, validation and trouble-shooting of complex bioreactor and down stream process processes would be required. This estimate could be increased to $150,000 and even that may be on the low end depending on where the factory is to be located. ”
There are a bunch of other critiques are basically arguing “this is expensive and nobody will fund it”, but that’s just an intuition, not a hard technical stop.
Here’s the layout provided. I’d love to see more of these, like the one WildType provides here