Correction: Originally the caption for Fig 3C and Fig 3D below were incorrect. Originally, the caption read that the top plot (Fig 3C) showed both Wuhan and Omicron-primed mice, when in fact it showed Wuhan-primed spike only. The red and black plots show different responses from epitopes derived from either the Wuhan spike or the Omicron spike. Apologies for the mix-up!
Okay, so I have tried many different ways to tackle this study. I was going to break it down figure by figure, but I found that was going to result in a post almost an hour long.
THEN, I was thinking about creating a post and asking questions for all of you to respond to. That would be great! However I got stuck on figuring out all of the questions/answers to asked, so that idea was abandoned as well.
So instead, I’m going to tackle this paper the best way that Modern Discontent can- with an incoherent rant!
Well, maybe not too incoherently (let’s say that coherence is relative…), but I’ll point out a few issues I have with this study. I also welcome different people’s opinions as well, including criticisms of my assessment or any criticisms about the study.
Why go over this study?
A few weeks ago I asked everyone what they thought about OAS/Immune Imprinting. I wasn’t just curious about people’s different opinions, but I’ve grown to become more concerned that people aren’t looking into some of the studies they are presented with. I think it’s becoming a bit too common for people to take what they see on Substack and trust that the information they are given is correct, even if the information has many faults (including the analyses). It’s even more concerning that many people may be hesitant to voice their own concerns and opinions that may run counter to the groupthink.
And so it may be fitting to just point out some of my criticisms about this study. I should make it clear that I generally don’t read studies with a fine-toothed comb, so for some this may seem hypocritical. However, it should remind us that all of us are not impervious to criticisms. If an analysis doesn’t seem quite right we should be allowed to criticize said analysis, as well as open to criticisms for ourselves.
A Critique of the Paper
Of course, at this point it should be known that the paper we are examining is this paper from Reynolds, et. al.1
At the time of this writing, it appears that this paper has been downloaded over 200k times, which is a pretty amazing feat to be quite honest.
As to why this paper was downloaded so many times, we’ll leave that for other people to speculate.
Keep in mind that many of these criticisms are things that I either find confusing are something that doesn’t quite make sense methodologically. It’s likely that some of these may be issues with my analysis (again, I welcome criticisms).
So let’s get to critiquing…
How do we define “imprinting”
This study heavily relies on the concept of immune imprinting, or OAS. This is one of the reasons why this study was cited by a few people- to indicate that this paper may be an example of OAS/immune imprinting from the vaccines.
The only problem is, that’s not how imprinting is defined in this study. Actually, it’s hard to figure out how the researchers define “imprinting”.
Generally, a paper attempts to define the terms they are working with before diving into their results- you’ll usually find these defined terms in the Introduction section, along with a good bit of background from prior research. It’s important that researchers define terms, so that even if terms are used outside of their normal use you can understand that the term is relative to whatever the study is.
However, when looking at this study I can’t make out how the researchers define imprinting.
For example, here’s a section from the Introduction:
First, following the earlier demonstration that people at this stage in the pandemic carry heterogeneous, immune-imprinted repertoires derived from their distinctive histories of infection and vaccination, we explored how these differences manifest in differential, cross-recognition of B.1.1.529 (Omicron) relative to other VOC, at the level of binding and neutralizing Ab, B cell and T cell immunity (24). Analyzing a London HCW cohort having detailed longitudinal, clinical, transcriptomic, and immunologic characterization, we considered the extent to which prior encounter with spike antigen through infection and vaccination shapes subsequent immunity to B.1.1.529 (Omicron) through immune imprinting.
Heterogenous makes sense (vaccine plus infection), but I still can’t make out what is intended to be implied by the use of “imprinting” here. If this all seems like a game of semantics, remember that the researchers are attempting to make a distinction between vaccination and infection (supposedly?). In doing so, we should at least argue whether vaccines are considered to be imprinting, or are we referring to infection alone. And if are only referring to infections, then what’s the difference between imprinting from Wuhan infection and no imprinting from a Wuhan-specific vaccine (if that makes sense- remember the problem is whether the Wuhan spike from either an infection or vaccine are treated the same or differently- this effects how “imprinting” is being used here)?
Honestly, we’re probably off to a bad start if we can’t even figure out what “imprinting” is referring to here (note that the last sentence refers to infection and vaccination in shaping immune imprinting).
No Natural Immunity Cohort
I’ve seen several people make comments/jokes that the massive vaccination campaign has essentially removed the control group (i.e. unvaccinated individuals). I took these comments as being tongue-in-cheek, but now I’m pretty sure this is the course for all studies moving forward; that essentially no natural immunity group will be considered in these types of studies.
Note in the Introduction excerpt above the researchers are making a bold assumption with their study- they’re assuming that nearly everyone has been either vaccinated or have been infected and vaccinated. Essentially, they’re making the assumption that those of us who did not get vaccinated but got infected alone shouldn’t exist (in the literature, at least), and therefore they’re not included in this study. In some regards, it makes sense considering the participants in this study were healthcare workers- there was no way unvaccinated healthcare workers would want to out themselves, and it’s likely mandates removed these participants anyways, thus heavily biasing their study in the first place.
This creates all sorts of problems when looking at these Immunology studies, in particular it doesn’t tell us anything about natural immunity, or if some form of immune imprinting may occur from those who are naturally infected- we just don’t have any data on that. And we’re likely to never get any data on that moving forward.
This also means that we can’t make assumptions that somehow those of us who are naturally infected are going to fare better than those who are vaccinated. We may have been able to make such inferences last year when we actually did receive evidence comparing natural immunity and vaccine-induced immunity2, but there’s no way such a study can be conducted now.
So how does all of this relate to this study?
It means that this study will only provide comparative analyses for those who are vaccinated and boosted- and nothing else. No, we can’t assume that those who were naturally infected with prior variants will not suffer from immune imprinting (based on this study), nor can we infer that an infection such as Delta then Omicron may confer better protection in the future, or any form of immune imprinting or OAS. We really can’t make any comparison to natural immunity at all. There’s just no way to make such an inference from this study alone, and it would likely be considered inappropriate to do such a thing.
This critique doesn’t refer to this study alone, but it refers to the literature which likely will be heavily biased away from any study on natural immunity, and thus we will have a huge gap in knowledge that we will likely not recover.
Imprecise Analyses
Brian Mowrey provided some criticisms of this study in his comment on Dr. Robert Malone’s post.
Most notably he made this remark:
Note also that the authors are serially imprecise or intentionally misleading whenever it comes to describing things as being “reduced,” never specifying “compared to what” on a case by case basis.
There’s a pretty egregious issue in that this study tends to make tenuous comparisons. The researchers seem to like to cite statistical significance, yet they seem to selectively choose when to include the magnitude to their values.
Take a look at one of the figures from this paper. In this portion of the study, the researchers took blood sera from HCWs, isolated the antibodies and challenged them against various spike RBDs (C) or in neutralization assays (D):
Those p-values above the box-and-whisker plots indicate whether the results are statistically significant (generally, p values below 0.05 are considered statistically significant, although the p value a researcher uses is an arbitrary value). The brackets underneath the p-values indicate which groups are being compared to one another. Given this information, we can see that many of the p-values above show statistical significance.
We also have this excerpt from the paper as well:
We found differences in immune imprinting indicating that those who were infected during the ancestral Wuhan Hu-1 wave showed a significantly reduced anti-RBD titer against B.1.351 (Beta), P.1 (Gamma) and B.1.1.529 (Omicron) compared to infection-naïve HCW (Fig. 1C). The hybrid immune groups that had experienced previous Wuhan Hu-1 and B.1.1.7 (Alpha) infection showed more potent nAb responses against Wuhan Hu-1, B.1.1.7 (Alpha) and B.1.617.2 (Delta) (Fig. 1D). However, cross-reactive S1 RBD IgG antibody and nAb IC50 against B.1.1.529 (Omicron) were significantly reduced compared to the other VOC irrespective of previous SARS-CoV-2 infection history (table S3 and Fig. 1, C and D).
So we’re told that the values above indicate significant reduction in anti-RBD titers, the only question is how much of a reduction is seen? The term significance here is a dangerous word- it’s likely referring to statistical significance (i.e. the decrease is “statistically significant”)- yet the researchers never make note about what the magnitude of this anti-RBD antibody reduction actually is3.
Arguably, if I look at the results from Fig. C for Beta VOC RBD column, the results for the uninfected, triple-vaccinated group (blue) looks a bit similar to the Wuhan-infected group (keep in mind this assessment of the plots would be considered highly improper, so I’m being somewhat facetious here), so unless we’re actually provided any value as to how much of a reduction has occurred, how are readers expected to make a comparison between the two groups? This is an issue in which scientists should not be ambiguous when it comes to their results, so for the researchers to be ambiguous about their actual values is rather strange.
This is made even worse when we look at the neutralization information (Fig D, on right). There’s hardly any information here, so why has there been no statistical analysis done? Is it because we may find out that prior exposure to other variants may, in fact, confer better neutralization compared to just vaccination alone?
I mean, if I looked at just the last column (neutralization against Omicron spike), I may infer that prior infection from either Wuhan or Alpha before vaccination actually provides better neutralization than just vaccination alone.
But we can’t have those results now, can we? Since those results would actually run against the idea that prior infections would result in immune imprinting. I’m just speculating here, but it seems rather improper for the researchers to have chosen their p-values a la carte. There really should be no picking and which values to report going on here.4
In short, there’s a general problem where the researchers here are selectively choosing which groups to compare, rather than comparing all of the results. This leaves readers to either interpret the study themselves, or to not perform their own interpretation.
Imprinting vs Priming
I’ll skip forward to the T-cell study that used transgenic mice (titled B.1.1.529 (Omicron) spike mutations encompass gain and loss of T cell epitopes). Transgenic means that the mice had human genes inserted into their genome. In this case, these mice were provided an HLA gene- DRB1 gene for the 04:01 allele to be precise. So remember my little rant from Sunday about HLA and diversity and whatnot when looking at this section of the study (it’ll be important later on).
These mice were given various peptides that were associated with epitopes from either the Wuhan strain of the spike or the Omicron strain:
The peptide pool containing B.1.1.529 (Omicron) specific S1 and S2 spike mutations and its ancestral Wuhan Hu-1 equivalent pool showed differential, sequence-specific T cell priming by either ancestral Wuhan Hu-1 or B.1.1.529 (Omicron) sequence specific peptide pools (Fig. 3A and table S5B). That is, immunizing HLAII transgenic mice with either ancestral Wuhan Hu-1 or B.1.1.529 (Omicron) sequence specific peptide pools allowed us to investigate differential, sequence-specific T cell priming that occurs as a consequence of B.1.1.529 (Omicron) spike mutations. We showed that priming with one pool resulted in impaired responses to the other (Fig. 3A).
That’s pretty interesting, so priming with one pool resulted in impaired responses against the other? Wait, I thought this was an imprinting study, not a priming study!
Okay, being a bit facetious again, but within the context of this study there’s a bit going on here that needs to be addressed.
That’s because priming isn’t the same as imprinting. In essence, priming suggests that the response to one stimuli changes the response to another, while imprinting would suggest some type of permanent or lasting effect after exposure to the first stimuli.
More importantly, priming essentially readies future responses to the same stimuli (note- same stimuli). T-cells start off naïve and must be presented with different epitopes, which then activates a cascade of responses that lead to cell differentiation in order to produce a quicker, stronger response to the same stimuli.5
With that, let’s look at the results from this section:
This is the graph that is referred to in the above excerpt, and as stated the response to one pool of spike epitopes somehow “impaired” the response to another.
Looking at this graph, we’re already running into a bit of a problem.
For one, note that there’s no control group- we’re led to believe that exposure to one pool impairs exposure to another pool, yet what response should we expect from one pool or the other? The researchers don’t provide any data on mice just provided one or the other to compare, so we don’t know what effect one exposure to Wuhan (or Omicron) would cause with respect to T-cell responses. Technically, using such a control would run counter to a study trying to look into priming of T-cells, but if we were to make such an argument we can also see why this study- as it is being presented- already has a ton of issues with it.
So this section doesn’t tell us much aside from the fact that, yes, some T-cells were primed (which is what we would expect).
But remember my quip about what epitopes are being targeted specifically (well, what antibodies are targeting which epitopes from my post on Sunday).
That leads me to the other two plots in this section. I think what’s more interesting is the plot for Fig B and C:
These plots provide a lot of context. It essentially took all of the peptide pools that the mice were exposed to and separated out the responses based on each individual epitope.
The top graph (B) is a mapping for mice exposed to Wuhan peptide pools, who’s T-cells were then exposed to individual epitopes from either Wuhan (red) spike or Omicron (black) spike. The bottom graph (C) shows mice who were exposed/immunized6 against Omicron and exposed to the same individual epitopes as in (B). Note that the mutations within the epitopes are shown at the bottom in an angled fashion.
Generally, what this tells us is that the T-cell response in Fig A boils down to only a few select epitopes, and that mutations within these epitopes alter T-cell responses.
For example, the DRB1*04:01 allele binds to the wildtype version of the Wuhan epitope shown in column 9 (Fig B), yet when an epitope carrying Q493R/G496S/Q498R/N501Y/Y505H mutations is used (as seen in Omicron), the HLA molecule no longer binds and responds to this mutated epitope.
This can be seen with the red plot disappearing below in between Wuhan and Omicron immunization:
And the opposite occurs with the A67V/del 69-70 mutation, such that the DRB1*04:01 HLA molecule does not bind to the wildtype version of this epitope, yet it binds to an epitope carrying the A67V/del 69-70 mutation.
This is all stated by the researchers:
We then looked at responses to individual HLA-DRB1*04:01 epitopes. Interestingly, while the B.1.1.529 (Omicron) mutations were associated in four instances with loss of a clear HLA-DR4-restricted T cell epitope (Fig. 3B: S371L/S373P/S375F, p = 0.0006; N440K/G446S, p = 0.0210; Q493R/G496S/Q498R/N501Y/Y505H, p = 0.0064; N679K/P681H, p = 0.0128), the mutated sequence epitopes in four instances led to de novo gain of Omicron-specific, HLA-DR4 T cell epitopes (Fig. 3C: A67V/del 69-70, p = 0.0152; G142D/del 143-5, p = 0.0152; Q493R/ G496S/Q498R/N501Y/Y505H; N679K/P681H, p = 0.0281; N764K, p = 0.0281).
So what’s the point of hashing this information if the researchers already commented on this gain/loss of epitope function?
Well, the problem resides with Fig A, and the supposed “impairment” of immunity that is going on here.
I actually have a lot of issues with the use of the word “impaired” here, since impaired would at least insinuate that the T-cells are not responding properly. However, there really is no way of figuring out whether or not any form of impairment has occurred above, and that’s mostly due to the interchange between the terms imprinting and priming.
Let’s conceptualize this using the comparing apples and oranges analogy.
Let’s say you are provided a bushel of various apples (let’s compare this bushel to the Wuhan peptide pool). You’ve never eaten apples before (naïve T cell), so you take a bite from a few different species (epitopes) and find that there are some you like and some you absolutely do not want to associate with. In particular, you find that apples such as Granny Smith and Macintosh apples are ones you prefer the most.
There’s also someone else who is ignorant of oranges who are then presented with oranges (Omicron peptide pools), and from there they choose Navel and Valencia oranges7(Omicron epitopes) as ones that they prefer.
In both instances these people have been “primed” i.e. they’ve been given either apples or oranges, and their preferences selected from different species (epitopes) of each. If we give someone who was primed with oranges some more oranges, they know which types of oranges they like the best.
So what if you give someone who’s primed with Oranges some Apples?
Well, they may have been primed to like Oranges, so we’re not sure what their response to Apples are. This would require that we give them apples, then give them apples again to see if they like apples, and which types of apples more specifically.
If we find that giving this person a few series of Apples doesn’t work- they still only prefer different types of Oranges, then we may think that the Oranges left some sort of lasting impression (i.e. imprinted).
But if we give someone who was first shown Oranges some Apples, should we assume these people would should have some innate preference for Apples?
This idea is a bit simplistic8, but this is one of the problems with the study above. The problem with this “impaired” immunity argument is that the researchers essentially gave these mice a bunch of different apples, then presented them with oranges, and now are complaining that they don’t like Navel or Valencia oranges without checking to see if they even tried those oranges.
And so we don’t know if the T-cell response is “impaired”- we are only assuming that to be the case without any verifiable proof.
Considering that the T-cell studies were somewhat based off of this finding, and was somewhat used to argue immune imprinting, it does raise a few questions as to the conclusions made in the other studies.
However, my interpretation and criticism could be way off. I’d like to know of other thoughts and criticisms (i.e. if I completely botched this analysis).
T-Cell immunity is too complex for the experiments provided
So I may have (actually I definitely did) belabored the point above, however this should probably tell us that an examination into T-cell immunity requires a lot of nuance, such as the inclusion of HLA haplotypes.
If we take the above experiment into account, which only looked at transgenic mice given the DRB1*04:01 allele, then could we consider what response other allele carriers should have? Certainly not, and that’s not even taking into account Class I molecules (which would be vital to those who were infected) along with other genes such as DQB1 and DPB1 genes for Class II HLA.
Because of all of these factors, it makes the assessment of the T-cell immunity portions of the study possibly moot. We don’t know the HLA haplotype of the HCWs, and we certainly don’t know what changes in spike epitopes between variants would lead to what changes in T-cell response. There’s just no evidence to make these conclusions, and it’s rather strange that the researchers included the transgenic mouse study yet didn’t take into account how narrow of a study that mouse study would be. There’s no way to compare the mouse study to the T-cell immunity study with the HCWs.
They explain just as much when commenting why they chose the DRB1*04:01 allele:
Due to the complexities inherent in mapping the effects of mutations in individual T cell epitopes across cohorts carrying heterogeneous HLA alleles, we mapped the differential recognition of the B.1.1.529 (Omicron) spike mutations using HLA-DRB1*04:01 transgenic mice (23, 24) (Fig. 3).
Because of that, I find it hard to consider the T-cell experiments without serious caveats and limitations.
What to do with this study?
There’s probably a lot more to say with this study, but I’ll save that for the comments for those who have questions/criticisms.
Overall, I didn’t expect to try and look too deeply into this study. However, the more I look at it the more I’m finding issues with it. There’s a good bit of contradictory evidence, and the researchers are selective in which analyses they report.
There’s also the issue of some of the sample sizes. Generally, this wouldn’t be an issue, but when you have only 3 samples and one seems like a complete outlier (Fig. 5A, Wuhan cohort) then you should be concerned how this would greatly affect your analysis.
I try to avoid doing this types of analyses because I generally don’t look too deeply into studies. I’m like to take a few studies based on their analysis without looking too deeply into the methodology.
However, I thought it rather fitting because of how widespread this paper has become, and how widely it’s been cited.
Because of all of these reasons, I just found it rather fitting to at least parse out some of the information.
With that being said, I’ll turn this over to you all. Did you find the study confusing, were there some things you thought were good or bad, or is there anything about my remarks/criticisms you find either confusing or wrong.
Please let me know below! I’d like to see what you all have to say!
If you enjoyed this post and other works please consider supporting me through a paid Substack subscription or through my Ko-fi. Any bit helps, and it encourages independent creators and journalists outside the mainstream.
Reynolds, C. J., Pade, C., Gibbons, J. M., Otter, A. D., Lin, K. M., Muñoz Sandoval, D., Pieper, F. P., Butler, D. K., Liu, S., Joy, G., Forooghi, N., Treibel, T. A., Manisty, C., Moon, J. C., COVIDsortium Investigators‡, COVIDsortium Immune Correlates Network‡, Semper, A., Brooks, T., McKnight, Á., Altmann, D. M., … Moon, J. C. (2022). Immune boosting by B.1.1.529 (Omicron) depends on previous SARS-CoV-2 exposure. Science (New York, N.Y.), eabq1841. Advance online publication. https://doi.org/10.1126/science.abq1841
Shenai, M. B., Rahme, R., & Noorchashm, H. (2021). Equivalency of Protection From Natural Immunity in COVID-19 Recovered Versus Fully Vaccinated Persons: A Systematic Review and Pooled Analysis. Cureus, 13(10), e19102. https://doi.org/10.7759/cureus.19102
As an example, take a look at the results from the Hachmann, et. al. study I referenced several times in prior posts and how the researchers included the magnitude within their figures:
Hachmann, N. P., Miller, J., Collier, A. Y., Ventura, J. D., Yu, J., Rowe, M., Bondzie, E. A., Powers, O., Surve, N., Hall, K., & Barouch, D. H. (2022). Neutralization Escape by SARS-CoV-2 Omicron Subvariants BA.2.12.1, BA.4, and BA.5. The New England journal of medicine, 10.1056/NEJMc2206576. Advance online publication. https://doi.org/10.1056/NEJMc2206576
I tried examining the Table S3 in the Supplemental Material to see if that was any indication about comparisons between cohorts. Unfortunately, the results here only compare one cohort to their neutralizing capabilities between Wuhan and Omicron (i.e. if we look at the blue group it compares how the blue group did in column 1 [Wuhan] and column 6 [Omicron], so there’s still no analysis there to interpret…
I try not to cite Wikipedia when possible, but here’s a bit from Wikipedia defining priming:
Priming is the first contact that antigen-specific T helper cell precursors have with an antigen. It is essential to the T helper cells' subsequent interaction with B cells to produce antibodies.[1] Priming of antigen-specific naive lymphocytes occurs when antigen is presented to them in immunogenic form (capable of inducing an immune response). Subsequently, the primed cells will differentiate either into effector cells or into memory cells that can mount stronger and faster response to second and upcoming immune challenges.[2]
https://en.wikipedia.org/wiki/Priming_(immunology)
There’s already a problem when these mice were primed with peptide pools rather than be provided Pfizer’s mRNA vaccine, so there’s also likely to be a difference in antigen presentation and response.
I may have had to look up different species of oranges…
I usually stay away from analogies because then it can get a bit too confusing and too far removed from the original intent.
So this comment is going to revolve around the supposed loss of T cell epitopes (Figure 3B/3C). I have spent an egregious amount of time giving the benefit of the doubt to these researchers, but alas, I cannot see how they have proven their result.
Just to recap the experiment (with my concerns in parenthesis):
1.) Mice were "immunized" with a pool of either Wuhan or Omicron spike peptides (now does injecting peptides into a mouse's footpad denote sterilizing immunity or just cellular debris to be collected by macrophages).
2.) After 10 days lymph nodes near the area of injection were removed and prepared as single cell suspensions (there are lots of immune cells in lymph nodes - many of which can secrete INFy - many of which are not T cells - there is no indication that any effort was made to separate out CD4+ cells which bind to MHC-Class II which I suppose is the whole reason for using transgenic DRB1*04:01 mice - however, CD8+T cells are the main source of INFy in lymph nodes and they bind MHC Class I - if you want to make a point about DRB1*04:01 they maybe you should only use a lymphocyte that binds Class II MHC - specifically CD4+ cells - there are ways to separate immune cells - magnetic beads or flow cytometry are routinely used to accomplish this).
3.) These cell suspensions were incubated with Wuhan and Omicron spike peptides - activated lymphocytes will secrete INFy which will bind to the biotinylated anti-INFy antibodies bound to the ELISpot plate. Then they add a streptavidin(binds to biotin)-alkaline phosphatase conjugate. And finally a BCIP/NTB which turns a pretty blue/purple in the presence of alkaline phosphatase.
The results certainly do indicate that mice primed with Wuhan are certainly having a reaction to Wuhan peptides. And, mice primed with Omicron are certainly having a reaction to Omicron peptides. That right there is called a secondary immune response. You can't immunize with one set of peptides and then challenge with the same plus another set of peptides - of course the first immunization will have a stronger signal - it's had more time develop. It would be more interesting if they did a Wuhan peptide challenge and after some time an Omicron peptide challenge and after some more time run this assay (but maybe with just CD4+ T cells). If only Wuhan or only Omicron signals were seen then it could suggest imprinting.
Also interesting (which they do not mention) is the T547K peptide clearly has a response for all challenges although it is not listed as a likely candidate for a DRB1*04:01 restricted epitope (Table S9). And making a big deal over the "loss" of the Q493R... DRB1*04:01 restricted epitope when the computer modeling again predicted its loss (umm, maybe the 3 amino acid difference had something to do with that). There's a lot going on in the paper and just accepting their results is a bit ludicrous.
On a lighter note: I enjoy Cara Cara oranges and Honeycrisp apples!
I'm probably one of those that downloaded this study, along with dozens and dozens of others I have also not read.
I only glanced at it, only thinking that this may actually be useful after we get to the worldwide recognition that the recipients of the "gene therapy" have damaged immune systems, politics be damned, and that this "(not a) vaccine" campaign will be viewed in the future as a bigger screw up than war, global warming, slavery, Communism, religious radicalism, centralized world government, or trans ideology.