Thread by @jbkrell

The @Dereklowe piece sort of misses the point. Yes, there is no idle mRNA vaccine production capacity in the pharma industry that could simply be repurposed to make more COVID-19 vaccines. But we could rapidly *build* more capacity, as we did across the world last year. 🧵

https://twitter.com/Dereklowe/status/1356665437505871872

Before we get deep into the weeds, lets remember, in January/Feb, the world had basically no commercial scale mRNA vaccine production capacity. By December, private industry (with a lot of public funds) built 3 billion+ doses / year scale capacity

So what happened? Companies like @pfizer, @BioNTech_Group, and @LonzaGroup (which makes the bulk of finished drug substance for @moderna_tx ) rapidly built mRNA vaccine production capacity inside existing manufacturing facilities.

To see how fast this can happen, let's look at @BioNTech_Group purchase of @Novartis biologics manufacturing plant in Marburg, Germany on 28 Sep 2020. (h/t @LeaozinhoM for this example) fiercepharma.com/manufacturing/…

Purchased at the end of Sept, with no built in mRNA vaccine production capacity. In Oct\Nov BNT installed equipment for mRNA vaccine production. In Dec the plant was approved by environmental authorities for operation. On 28 Jan 2021, the plant was approved for cGMP production.

The plant starts commercial production this month. So a very new German company, in less than 6 months, was able to convert an existing plant w/ no mRNA vaccine production capacity into a plant that can produce 750 million doses per year.

Almost certainly if an entity like the USG was involved, this process could go faster (albeit not much faster), because of basically unlimited capitol, labor and special powers like the DPA and 28 USC 1948.

Now lets talk about the step that @Dereklowe argues is the rate limiting one, lipid nanoparticle (LNP) encapsulation. The LNP are the tiny (prob. ~100 nm), roughly spherical particles made of lipid (a fat is a lipid) that enclose the modRNA payload.

The LNP is incredibly important for mRNA vaccine efficacy. It allows the RNA to get into the inside of the cell, allowing the RNA to be converted (translated) into the (modified) spike protein that elicits the immune response. Obviously simplifying a lot here.

Quality control is super important here, in vitro results show that tiny differences in LNP composition/size/morphology - may have big impacts on
efficacy. So we need to make sure that the LNPs generated are within narrow set of parameters. e.g sciencedirect.com/science/articl…

The process of generating LNPs is called "nanoprecipitation" and generally involves precisely mixing the lipids (dissolved in semi-polar solvent, like EtOH) with the RNA (dissolved in a buffered polar solvent). The dynamics of this mixing process must be carefully controlled.

@DerekLowe assumes that LNP generation for vaccines uses microfluidics to generate the LNPs. This is a reasonable assumption, but not necessarily correct. Recall that there are macrofluidic processes that can generate RNA containing LNPs.

For example, T-junction mixing is commonly thought to be able to produce RNA containing LNPs with pretty precise size control and good encapsulation efficiency. E,g onlinelibrary.wiley.com/doi/10.1002/sm….

In fact, @pfizer itself seems to imply that their production process uses a macrofluidic process for LNP generation. pfe-pfizercom-d8-prod.s3.amazonaws.com/Vaccines_Infog…

But the assumption that both vaccines uses a microfluidic process for at least part of the nanopreciptation process is reasonable and probably correct. But so what? There is a certain amount of hand waving going on here, like it uses microfluidics ergo we can't scale it up.

@Dereklowe says that these microfluidic nanopreciptation macihnes are not lying around in existing pharma plants, therefore we can't scale it up. True, but they weren't lying around @LonzaGroup, @pfizer plants in February either. But we were able to rapidly *build* that capcity

The manufacturing processes used in the production of most microfluidic "chips" are relatively easy to scale up. Recall that most microfluidic devices for LNP generation use soft lithography for manufacturing using e.g. PDMS for molding. e.g sciencedirect.com/science/articl…

This is *similar* to the process used to fab semiconductors. And recall that the feature size of these chips is generally way above 1 micron, we can use standard transparencies for the fabrication process. future-science.com/doi/10.2144/05…

Tl;dr the world scaled mRNA vaccine production capacity by more than 6 orders of magnitude in less than a year. there is nothing that stopping us from scaling up more now, except our will.