Here’s a good article from the Washington Post that updates some details that I talked about in this post on the lipids used in the mRNA vaccines and in this one about their overall manufacturing process. It focuses on Acuitas, who make the two proprietary lipids that are used in the Pfizer/BioNTech vaccine. Actually, if you go back a step, you get to Avanti Polar Lipids down in Alabaster, AL (as mentioned in Jonas Neubert’s terrific dive into the supply chains for these vaccines). Update: fixed some hosed-up links.
As you would imagine, everyone involved is having quite a time meeting demand. As the article says, though, it’s hard to get a clear picture of where the exact pinch points are, because the terms of the various contracts are not public, nor are even the identities of the raw material suppliers further up the chain. But anyone who’s had to source intermediates or starting materials in the drug business at any kind of scale will appreciate that the fastest way to find out what those weak points are is to place a Big Ol’ Order for something.
Among the interesting things you can uncover with that experiment is that what looks like several suppliers for Important Ingredient X are nothing of the kind, because they’re all (surprise!) sourcing it from the exact same place, likely as not somewhere back in China. That effect definitely happens with the Chinese suppliers themselves – a list of supposed sources across southern China will turn out, on closer inspection, to resolve to a single shop out in Chengdu or (even more worryingly) a bunch of drums in a warehouse near Shanghai from a big manufacturing overrun a few years back. The general tendency of many suppliers in every country to give you quotes for things that they don’t quite have yet, but are pretty sure they can round up if you place an actual order, also has to be taken into account.
In this case, you can be sure that everyone got down to the proverbial brass tacks very quickly, but it seems clear that there have been some difficulties sourcing starting materials on the scale needed and/or running some of the chemistry. These lipids involve multistep syntheses, and that’s just going to chew up time no matter what. This article at C&E News will give you some more details on that process, and I especially like this part:
Matthieu Giraud, global director of CordenPharma’s peptides, lipids, and carbohydrates business, says the synthesis requires about 10 steps and several product isolations. A complete manufacturing campaign is measured in months.
That does sound delightful. That team gets eaten up by all kinds of things: what sorts of purification are needed before the next step can be run, what the batch-versus-continuous-processing landscape is like, what scale the chemistry can be done at in the first place, and how that matches up with the sheer physical capacity of the plants doing the work. That scaling question alone is the door to a world of headaches: maybe one of the steps has an exotherm in its chemistry, which can mean special mixing conditions and size limitations. Or the product is especially thick and viscous and is thus harder to transfer and purify. There might be something tricky about the addition of a particular reagent – it’s corrosive unless you use certain alloys, or has to be warmed up to flow reliably and not all your reactors have that capability, or it doesn’t mix easily when it goes into the bulk reaction, or maybe something is theoretically available from several suppliers, but only one of them can furnish it in the sort of pellets that work best on scale. It goes on and on.
The topic of mixing brings up the formulation question that is one of the key steps in the whole process, forming the lipid nanoparticles with the appropriate amount of mRNA in each. I found this part quite interesting:
Companies have had to build equipment from scratch, including machines that shoot two streams of solution — one containing mRNA and one containing lipids — into a high-speed collision to fuse the nanoparticles and encapsulate the genetic payload.
Exactly the sort of thing I was picturing here. No, this is a pretty weird process compared to most production lines. There are two ways to look at this whole business, and they both have validity. The first is “Why the hell didn’t we anticipate this? Why didn’t we put more money into the raw materials supply chain and the manufacturing capacity, and plan for success?” I’m sure that there are parts of this that could have been done better, but at the same time I’m also sure that Acuitas and the other players in this area have been scrambling since early last year to try to meet a vastly scaled-up demand as well (that C&E News story above confirms this). Some of these problems could have been solved or at least ameliorated by throwing money at them, and some couldn’t – and remember, buckets of money were in fact showered on every part of the process. But even more could quite possibly have helped.
The second viewpoint is to be struck by how much we’ve been able to scale up these things anyway. That might sound too happy and rosy, but there’s something to this take as well. When you look at that WaPo article, one thing that hits you is that the Acuitas people and others who have been involved in this area for a long time are stunned by how far it’s come and how quickly. I’d be willing to bet that if you’d called any of them up in (say) December 2019 and asked them if they could get to where they actually are by February 2021 they’d have been terrified.
My guess is that the truth is in between those two: I certainly doubt that the mRNA scaleup process has been flawless, but I don’t think that people have exactly been bumbling around, either. I continue to be thrilled that the vaccines work, and work as well as they do, and I will be rolling up my sleeve for them the first chance I get.