The pre-2020 BNT162 development program

Acuitas & BioNTech were waiting for Godot since at least February 2019

A closer reading (infuriatingly many months later) of the first animal study on PHMPT, R-20-0072, shows that the BNT162 development program has been running at least since 22nd February 2019. This is the production date of a first batch of luciferase mRNA, FK190222-01c, visible in an improperly censored table in the Certificates of Analysis section.

The Certificate of Analysis of the second batch, RNA-EH190611-01c, clarifies that the six-digit number in the batch name is the date of production; assuming “date of order” refers to the beginning of production in turn syncs well with the Agilent quality control three days later.

In other words, BioNTech were making luciferase mRNAs and sending them to Acuitas to be formulated in different LNP mixtures, looking for the perfect delivery platform for systemic antigen expression creating many antibodies and thus being a good vaccine from the beginning of 2019. See the two different Acuitas delivery sheets below.

The November 2019 summary sheet and it’s two formulations corresponds well with the “pilot formulation” study mentioned in the “Annex 1 - Draft 3.2.P.2.2 Drug Product” document from the EMA leak:

The Acuitas R&D collaboration process is also described in detail (yet without dates). The December summary sheet, which we can assume to list LNP5 (ALC-0218), BioNTech’s DODMA:DOPE nanoparticle platform, and control (or perhaps rather the “properly” censored second item in the Physicochemical characterization table), is most likely the R-20-0072 study material.

The TGA Non-clinical evaluation (page 40) contains a lot more information on R-20-0072 than the censored PHMPT study report itself. However, while many elements of the TGA copy and the study report coincide, there are very peculiar differences between them.

When you compare the blue, green, and purple highlighted areas in the TGA and study parts, you can see several errors in the TGA version: wrong luciferase and lipid batch for DODMA:DOPE, correct Luc/lipid batches for LNP5/8, none of the technical measurements are what Acuitas reported. Unsure of significance, yet remarkable. Hypothetically, there might be differences between Acuitas’ and BioNTech’s measurement techniques/devices/calibrations, or TGA personnel writing the report mixed up analysis certificates from different studies. The other markings are as follows: blue box shows that the experimenter’s initials are in the lipid batch number for DODMA:DOPE, cementing the notion that this was non-Acuitas material (reading the study, they explicitly say so, but nevermind), red box shows that DODMA:DOPE was the third test article in R-20-0072 and where the TGA “LNPC12” name might originate, orange box is the second batch of luciferase.

Interestingly enough, the DODMA:DOPE batch used FK190222, thus BioNTech should have included the corresponding certificate of analysis from February in the appendices. Did other regulators make the same “error” assigning the second luciferase batch to the DODMA:DOPE test article of R-20-0072, thus absolving BioNTech of having to add a rather embarrasing February 2019 mRNA order sheet to the study material? May 2019 is bad enough, to be honest, it’s probably just a mistake. Probably.

Another point of uncertainty was the FK prefix of the first batch, yet other BioNTech batches (before they changed their internal batch naming process to not reflect the completion date) show the same structure: two letters+six digits+extraneous characters. The second luciferase batch has “EH”, the first BNT162b1 batch has “KG”, etc.

What remains slightly nebulous is the significance of “mod HA” compared to “mod Luc”. The former assuredly codes luciferase, as evidenced by the uncensored bioluminescence imaging in the TGA report.

Transposing the 16 days between date of order and the date in the batch name onto the FK batch gives a date of order of 6th February 2019. A year later to the day, Acuitas boss Thomas Madden would be in Germany, discussing the BNT162-01 clinical trial with BioNTech and German regulator Paul-Ehrlich-Institut. Here is an excerpt from a December 2020 PharmaBoardroom interview:

BioNTech had obviously realized this far earlier than “when the pandemic hit”, as evidenced by producing pseudouridylated luciferase mRNA as early as February 2019.

My impression: the pilot formulation + R-20-0072 studies were a set piece. BioNTech had entered a flu vaccine collaboration with Pfizer in July 2018, a perfect cover to start pivoting from personalized cancer treatment to infectious diseases. As evidenced by the March 17 2020 agreement between Pfizer and BioNTech, Pfizer was not privy to the Acuitas R&D. This is further supported by the first mention of Acuitas in BioNTech’s SEC filings being in 2021, where BioNTech discloses licensing Acuitas technology. BioNTech was “developing” an infectious disease vaccine platform with Acuitas on the low, spending a lot of money for the technology when they’re already partnered with the biggest pharma company in the world?

I say “developing” due to the auspicious timing: luciferase batch FK190222 is from february, EH190611 is from June, yet the first delivery of product to BioNTech was 26th November. The next delivery sheet is only fifteen days later. Fifteen days in which BioNTech does the pilot formulation study, evaluates the results, and Acuitas formulates the products for the December 11th delivery sheet and study R-20-0072.

That’s not development, that’s a scripted setup that waiting to be enacted. German chancellor Angela Merkel’s visit to Wuhan in September 2019 certainly takes on a whole new light.

A quote from EMA leak Annex 1:

Formulation development began with screening of lipid nanoparticles. This involved incorporation of the novel lipids into LNPs encapsulating RNA and then submission to a screening process which included pilot formulation and physical characterization to determine suitability for in vivo testing and activity determination in a murine model of RNA expression.

During pilot formulation, candidate cationic lipids were formulated with RNA encoding for luciferase and other component lipids in a standardized composition which comprised the cationic lipid (ALC-0315, ALC-0218, etc.), a PEGylated lipid (ALC-0159), DSPC and cholesterol. The use of RNA coding for luciferase allowed for visual assessment of expression and translation in model systems. Results of these studies led to the selection of ALC-0315.

Just terrifying how all regulating bodies casually overlooked the dates on these studies and research developments.

There are one or two further luciferase-LNP8 animal studies between R-20-0072 and 38166 on the non-clinical development timeline involving ApoE knockout mice which are summarized in the Drug Product annex.

Another very interesting smidgen of information from the EMA leak (quality report p56) is another hint concerning the differences between BNT162b2v8 and v9: a footnote in a diagram states that “RBP020.1 and RBP020.2 code for the same antigen, but with variations in the used codons, thus, they differ slightly in the nucleotide sequence. Codon optimization has been performed either via a BioNTech proprietary algorithm (RBP020.1) or with a published algorithm (Raab et al. 2010; RBP020.2).”

RBP020.1 is b2 v8, RBP020.2 is b2 v9, which ended up being Comirnaty. v9 elicited 3-5x more antibodies than v8. BioNTech must suck if their proprietary codon-optimization got absolutely wrecked by an open-source program. Absolute shenanigans when you think about it: they exploited their own “incompetence” to get well codon-optimised modRNA into humans without a prior tox study; 38166 used BNT162b2v8, and the repeat-dose tox study for BNT162b2v9, 20256434, started the same day Phase 3 dosing in C4591001 began. Did they codon-optimize BNT162b2v8 “wrong” on purpose, in order to have the consensus optimal sequence be much better and skip the line into human trials?

I’m still not entirely sure about the Genevant/Arbutus/Acuitas patent love triangle. Perhaps BioNTech got a taste of the ALC formulations in their collaboration with Genevant and decided to go straight to the source? Which is why Vivek Ramaswamy is suing them, I reckon. The July 2018 flu collab with Pfizer seems as if Pfizer could smell a windfall looming, but BioNTech played a fast one on them and forced them to negotiate a new agreement in March 2020. All purely speculative, of course.

There’s also the issue of tech transfer to Polymun and Dermapharm from Acuitas, which had to have been complete in time for the anachronistic 26th March certificates of analysis in 38166, but that’s for another article.