Unfortunately, I have found myself writing a lot about the antivax false claim that COVID-19 vaccines are causing a wave of “turbo cancer” in young people, mainly because, aside from the usual quacks like Peter McCullough, Harvey Risch, and William Makis, a seemingly “respectable” cancer doctor named Wafik El-Deiry has also been amplifying such claims in the name of “open-mindedness” and wanting to consider “all scientific possibilities.” In brief, the claim is that COVID-19 vaccines are causing “turbo cancer,” which is never really well-defined as a term but seems to mean—in antivax parlance, at least—rapidly growing and unusually aggressive cancers arising de novo or the recurrence of cancers that had been in remission in a particularly aggressive and fast-growing form soon after vaccination with mRNA-based COVID-19 vaccines. The fantastical mechanisms proposed are many, but most come down to the claim that the vaccines are “contaminated” with plasmid DNA used to make the mRNA used I the vaccines and left over from the manufacturing process and that that plasmid DNA contains SV40 sequences. It’s a conspiracy theory that goes right back to the old conspiracy theory that contamination with SV40 virus found in polio vaccines in the early 1960s was responsible for a wave of cancer decades later. (Hint: It wasn’t. Also, SV40 promoter sequences are not the same thing as the intact SV40 virus, the latter of which is oncogenic.)
After my most recent post on “turbo cancer” claims, I thought I was done for a while. Unfortunately Wafik El-Deiry got all credulous again and decided to amplify another paper by Kevin McKernan as “concerning” that there might be something to this whole thing about plasmid DNA contamination in COVID-19 vaccines and “turbo cancer”:
The preprint being flagged as “concerning” is apparently McKernan’s followup to his original awful study of his that I first discussed in depth a few months ago and that led to followup studies that were equally bad by scientists like Phillip Buckhaults. I will agree that the study is “concerning,” but not because it provides support for the claim that mRNA vaccines are massively contaminated with plasmid DNA to the point that they can cause “turbo cancer,” Prof. El-Deiry’s post/Tweet notwithstanding. A quick read of the preprint, DNA fragments detected in monovalent and bivalent Pfizer/BioNTech and Moderna modRNA COVID-19 vaccines from Ontario, Canada: Exploratory dose response relationship with serious adverse events, revealed many problems to me other than its having been coauthored by at least two antivaxxers, Kevin McKernan and Jessica Rose and abusing the Vaccine Adverse Events Reporting System (VAERS) database in a way that antivaxxers love to do.
The CliffsNotes version of the paper’s findings is that there is a horrible amount of plasmid DNA contamination in mRNA-based COVID-19 vaccines manufactured by Pfizer and Moderna and that the amount of plasmid DNA in the vaccines correlates to an increased incidence of adverse events (AEs). If you take what McKernan and colleagues have done at face value, they claim to have found more VAERS AE reports for COVID-19 vaccine bearing lot numbers that, by their analysis, had higher levels of plasmid DNA “contamination’ compared to vaccine lot numbers with lower levels of plasmid DNA “contamination.” I bet you can already see the problem here, but we’ll get into the issues with the study in more detail. I also note that, for unclear reasons, McKernan used two different methods to assay DNA “contamination” in the vaccine vials, the gold standard of quantitative PCR and an assay based on the binding of a fluorescent protein to the DNA and RNA called the Qubit assay. Depending upon the specific fluorophore used, a scientist can distinguish between double-stranded DNA (like plasmid DNA), single-stranded DNA, and RNA. (mRNA is single-stranded.)
Before I dig in, though, let me just cite Ed Nirenberg, who has pointed out that we literally have DNA vaccines that use orders of magnitude more intact DNA of much longer length than any amount of DNA you could possibly find in the Pfizer and Moderna COVID-19 vaccines:
As an aside, I know just how correct Ed is because three decades ago I did my PhD in a lab where one of the projects involved getting muscle to express proteins coded by plasmid DNA and to use that technique to see how gene regulatory sequences functioned in vivo rather than in the very artificial conditions of cultured cells. (Here’s a representative paper from that lab.) I didn’t work on the project, but we did have lab meetings in which each of us would present our work; so I did see lots of presentations on the progress of the research. Also, DNA is supposed to be in the nucleus, not the cytoplasm, which is another reason why there are mechanisms in cells to get rid of it when it is detected in the cytoplasm:
In fairness, at least with this study we are told where the vaccines used in the study came from in more concrete terms:
Expired unopened vials of Pfizer-BioNTech BNT162b2 (n=8) and Moderna Spikevax 113 mRNA-1273 (n=16) were obtained from various pharmacies in Ontario, Canada (Figure 114 1). Three vials of in-date remnants of the same lot of Moderna XBB.1.5 vaccine were also 115 obtained. In total, 12 lots were surveyed across 27 mRNA vials: 5 lots of Moderna 116 child/adult monovalent, 1 lot of Moderna adult bivalent BA.4/5, 1 lot of Moderna child/adult 117 bivalent Wuhan-BA.1, 1 lot of Moderna XBB.1.5 monovalent, 3 lots of Pfizer adult 118 monovalent, and 1 lot of Pfizer adult bivalent Wuhan-BA.4/5 vaccines. An unopened 119 sterile injectable vial of alprostadil 66 mcg/mL in combination with papaverine 21.7mg/mL 120 and phentolamine 1 mg/mL (TriMix) was used as the negative control. The unopened 121 vials were untampered as they had intact flip-off plastic caps with printed lot numbers and 122 expiration dates. Vials had been stored in a purpose-built vaccine unit at +2-8oC in the 123 pharmacies and were transported in insulated containers with frozen gel packs and 124 placed in the testing laboratory fridge within 5 hours. Only one Moderna vial did not have 125 a printed expiration date but had a QR code that required scanning by a pharmacist. The 126 Moderna XBB.1.5 vials were similarly stored by the pharmacy. Vials were removed from 127 the refrigerator, warmed for ~20 minutes, and administered by the pharmacist to patients 128 over ~30 minutes. The remnant vials were placed in an insulated container with frozen 129 gel packs and transported to the testing laboratory fridge within 12 hours. 130
Samples from these vaccines were subjected to quantitative PCR (qPCR) using sets of primers designed to amplify sequences associated with the gene for spike protein, ori (the origin site of the plasmid used, presumably a measure of just the plasmid), and the SV40 promoter-enhancer-ori DNA. A standard curve was constructed using known amounts of target DNA, so that the signal obtained from the unknowns could be quantified and the amount of target DNA in them calculated. As I’ve explained before, the SV40 promoter-enhancer is often used in plasmids because it is a very strong promoter that can drive the production of lots of the desired mRNA encoded by the cDNA sequence attached to it. (Seriously, it’s nothing more nefarious than that.) Even though the plasmids used for mRNA vaccines are grown in bacteria and then transcribed to mRNA in an in vitro reaction, a lot of “stock” plasmids have SV40 promoter sequences in them. In this case, although I don’t know the reason why SV40 is included on this particular plasmid, I suspect it’s for dual use purposes and possibly as part of the attempted development of a DNA vaccine, so that the same plasmid could be used to generate spike protein in cell culture. Whatever the reason, it’s not dangerous.
Here’s a map of a typical generic plasmid, just to give you a visual reference:
For the Qubit assay, the authors used AccuGreen® HS fluorescence reagents from from Biotium. The authors also did some assays to determine the distribution of sizes of the DNA fragments in base pairs (bp). Let’s start with their “money” graph, Figure 5:
Does anyone notice anything about the top panel of the graph? You should. Basically, by the authors’ own measurements, the amount of DNA/vial fell below the FDA guidance of 10 ng DNA/dose. Did you also notice the little trick they did? They used a log scale to make the total DNA appear to be much closer to the FDA-recommended limit than it really is. For instance, in all the Moderna vials, the amount of DNA isn’t half of the recommended limit, it’s less than one-tenth the recommended limit, and, in the case of the ori sequence, well under 1/100 of the limit. The abstract itself even notes that the authors found DNA at “0.28 – 4.27 ng/dose and 0.22 – 2.43 ng/dose (Pfizer), and 0.01 -0.34 ng/dose and 0.25 – 0.78 ng/dose (Moderna), for ori and spike respectively measured by qPCR.” So, from McKernan and Rose’s own data, the vial with the very highest concentration of DNA was one Pfizer vial that had less than one-half the maximum DNA amount recommended by the FDA, while the Moderna vial with the most plasmid DNA contamination had less than one-tenth the maximum recommended by the FDA. In other words, there’s a whole lot of nothing here so far.
Before I move on to the Qubit assay results, I note that there’s also an issue that I wondered immediately about regarding why there should be an order of magnitude less ori sequence in the Moderna vaccine compared to spike sequence. In the Pfizer vaccine, there were roughly equal amounts of ori and spike is what one would expect, given that each plasmid only contains one cDNA for spike protein. It’s very odd and makes me question the validity of the assay. Certainly, McKernan and Rose never try to explain this discrepancy. If I were a reviewer for this article, I would certainly bring up that observation and ask for an explanation. (In fairness, the authors do mention it, but do not even speculate why they observed what they did for the Moderna vaccine.)
So now what? The authors got a result that they didn’t like using qPCR, specifically residual DNA content well below (and, for Moderna, quite well below the maximum recommended by FDA guidelines). What to do next? I think you know the answer. They used a different assay to do the same measurements. Enter Qubit.
By Qubit, they measured much, much more residual DNA (presumably plasmid DNA) in the vaccines, orders of magnitude more:
The amount of residual DNA varied substantially between lots (0.28 – 4.27 ng/dose for Pfizer ori, 0.22 – 2.43 ng/dose for Pfizer spike, 0.01 – 0.34 ng/dose for Moderna ori, 0.25-0.78 ng/dose for Moderna spike) when tested by qPCR. Fluorometer based measurements (e.g., Qubit®) of the vaccines show 2,567 ± 618 ng/dose (range: 1,896 to 3,720 ng/dose) for Pfizer and 4,280 ± 593 ng/dose (range: 3,270 to 5,100 ng/dose) for Moderna suggesting a high fraction of the DNA is under the size range of the qPCR amplicons.
A rule of thumb if you’re a scientist: If two different analytical methods result in such different readings, something is going on and that something is usually method-related rather than a real difference. Of course, note how, instead of investigating further to try to figure out the reason for the discrepancy, McKernan and Rose assume that they are seeing a real difference and immediately look for a reason to conclude that the DNA measurements orders of magnitude higher are the “correct” ones that most reflect reality. Elsewhere, they try to argue that even very tiny DNA fragments are highly dangerous:
The FDA guidelines are also written to only quantitate DNA fragments of 200 bp or 518 greater, in part because fragments smaller than this were not considered to be able to 519 produce a functional gene. However, Klinman et al.,31 suggests that fragments as small 520 as 7bp can pose integration risks.
Naturally, I looked up the reference cited, which notes:
In evaluating the potential harm of plasmid integration, it should be noted that the risk of introducing plasmids with strong regulatory regions into the host genome far exceeds that associated with random point mutations [43;50]. Moreover, the technology used to detect plasmid persistence does not examine the frequency with which short fragments of plasmid integrate. In this context, sections of DNA as short as 7 bp can affect rates of integration or recombination. Examples include the VDJ recombination signal sequence and related sequences, chi-like elements and minisatellites, ALU sequences, a recombinase signal present in hepatitis B and mammalian genomes, and topoisomerase II recognition sites .
This section appears to me to be talking not about short sequences on an intact plasmid that serve as regulatory elements in the context of DNA vaccines, which would involve the injection of huge quantities of DNA. It’s a different beast, and using that reference is rather deceptive.
In fairness, though, the authors did do one thing, which was to do use Oxford Nanopore sequencing to quantify the average size of the DNA fragments detected, but they only did it with one “previously studied vial” of Pfizer vaccine. Naturally, their results don’t support their speculation that there are huge quantities of short bits of DNA in the vaccines not being detected by qPCR. In fact, the amount of DNA between 45 and 73 bp is very tiny.
This brings up another issue. The amplicons (stretches of DNA amplified by the PCR primers used) are only a little more than 100 bp (105 bp for ori, 114 bp for spike). Only a small percentage of DNA is less than 100 bp. Also, as I like to point out (and would have, if I were reviewing the manuscript), you can amplify shorter sequences, although in fairness it’s hard to amplify much shorter than around 70 bp, and for qPCR an amplicon size between 70-200 bp is generally recommended. Still, think about what McKernan and Rose are saying. They are claiming that, by Qubit analysis, there are greater than microgram quantities of residual plasmid DNA in the vaccines, which is a huge amount of DNA. By Qubit, the average amount of DNA is 600-fold higher than the maximum measured by qPCR in Pfizer vaccines; for Moderna vaccines, it’s nearly 5,500-fold higher. These are discrepancies of a magnitude that would make any good scientists search hard for a methodological explanation rather than just accept the highest number because it’s the most useful for their narrative, the latter of which is what, of course, McKernan and Rose did.
Then they claimed that they could explain the higher number by Qubit by their supposedly being huge quantities of short DNA fragments that PCR won’t detect. Yet their own Nanopore analysis, albeit of only one vial, does not support this speculation for the cause of the discrepancy between the two methods. I also had another thought. I know from reading the package inserts (irony of my noting this understood) that the AccuGreen fluoresecent dye used is “highly selective” for double-stranded DNA (dsDNA) over RNA. However, one would expect, given the qPCR results, that the vaccine, once the lipid nanoparticles (LNPs) were disrupted by 95° C heat before PCR and before the Qubit assay, would have orders of magnitude more RNA than dsDNA. When there is a lot more RNA than dsDNA, even a highly selective assay could be affected by the RNA.
Yet McKernan and Rose didn’t do the most obvious control: Treat the solution with RNase, an enzyme that chews up RNA but not DNA, before doing the Qubit assay. (A qPCR assay wouldn’t be affected by a lot of RNA because the RNA was not reverse-transcribed.) Were I reviewing this paper, I would have insisted on that control. I mean, the authors used DNase to chew up DNA not in LNPs in order to estimate how much DNA was in the LNPs (and thus protected from DNase digestion) than was free in solution wit the LNPs. Surely they could have used an RNase to make sure that the large amounts of RNA one would expect in the LNPs was not interfering with their fluorescence assay. I mean, it’s an obvious control that they didn’t do.
This brings us to the VAERS results, which are hilariously bad. Rose and McKernan looked at the lot numbers for each vial and then searched VAERS for AE reports for that lot. Then they divided the number of serious AEs by total number of AEs for each lot and then graphed them against the DNA concentration that they had measured in the vials they had from that lot.
Figures 8 and 9 tell the tale:
Figure 9 shows the qPCR results for Pfizer (upper panel) and Moderna (lower panel). The Moderna results are plotted in separate panels because of the huge differences found between ori and spike sequences. First of all, these graphs are what we used to like to call “star charts” in the labs that I’ve worked in. There are too few points for anything resembling statistical power, and there doesn’t appear to be much of a correlation. Even more hilarious, look how they did their curve fitting:
The curves were plotted on a logarithmic axis and a trend line drawn using the 227 linear function within Microsoft® Excel.
You read that right. They used Microsoft Excel to do the curve fitting! Seriously, though, they could have used any software, and the results would have been just as risibly bad. Their using Excel instead of a real statistics package is just the chef’s kiss, the cherry on top of the crap science sundae, particularly given, as Ed Nirenberg points out, the Qubit assay actually contradicts their hypothesis:
Meanwhile, someone else also noted:
I’ll be honest. By this point, I was too tired to go dumpster diving into VAERS to try to figure out how much Rose and McKernan had gotten wrong. It wasn’t necessary anyway. Even if every single VAERS entry they cited were absolutely perfectly cited, their analysis is still laughably awful science so bad that I’ll assume that Prof. El-Deiry never bothered to read the methods or the rest of the manuscript aside from the abstract and what McKernaan said about it on Twitter, like this:
In the end, this is yet another awful study that, even if taken at face value and assuming the methodology was valid and analyzed correctly, doesn’t show what the authors claim that it shows. Add the methodological questions and the lack of some key controls, and it’s not only not good evidence for harm from residual plasmid DNA left in COVID-19 vaccines, but it’s downright uninterpretable.
With this study, McKernan and Rosen have joined the long, dishonorable list of antivax “scientists” producing bad science to support their antivaccine fear mongring going all the way back to Andrew Wakefield.