We all know that the Omicron variant has been a game changer: the various omicron subvariants become more and more transmissible, but have less tendency to induce severe disease, although elderly, people with co-morbidities and immune-compromised patients remain vulnerable.
From the perspective of vaccine protection, it has been clearly demonstrated that omicron is a master in immune escape and that “imprinting” may jeopardize our efforts to hit the “moving target”. The concept is that repeated vaccination with the original Spike (in whatever format) would only stimulate immune responses to already escaped epitopes and, importantly, divert the immune system from the new epitopes.
To try and overcome this phenomenon, “adapted” vaccines have been distributed, based on either BA.1 (more used in Europe) or BA.4/5 (more used in US). However, there are two problems:
- By the time these vaccines were rolled out, the virus had evolved further towards BQ and later XBB variants, which are even more immune escaping
- Both vaccines were formulated as “bivalent” = combination of Wild Type (WT = D614G) Spike with either BA.1 or BA.4/5 Spike, with as a consequence that “imprinting” by WT (and possible diversion from Omicron) is probably “built in”.
In this Episode, we will first discuss three studies, which claim that additional booster vaccination may actually lead to more instead of less infections during the omicron waves. In Par 2, however, we review the many studies that still show favorable effects of booster vaccination, especially against severe disease. In Par 3, we will discuss some immunological aspects that may help to understand and possibly resolve the issues on imprinting.
I hope this Episode will promote some discussion and exchange to clarify our view on these complex phenomena.
Par 1 Three studies suggesting that a higher number of (mRNA) vaccines increases the risk of infection during several Omicron subvariant waves
Ep 332-1: Eythorsson JAMA Network Open Aug 2022: Risk of re-infection in a national cohort of previously infected subjects Dec 2021-Feb 2022 (= Omicron BA.1/BA.2) in Iceland
As can be seen: re-infection risk higher:
- in young adults (18-29)
- longer time period since initial infection
- 2 doses or more of vaccine as compared to 1 dose or less (= unvaccinated)
Discussion by authors: This finding should be interpreted with caution because of limitations of our study, which include the inability to adjust for the complex relationships among prior infection, vaccine eligibility, and underlying conditions.
Remark: no data on severity of re-infection.
Ep 332-2: Haim Chemaitelly medRxiv Nov 2022: Matched, retrospective, cohort study of re-infection in the national cohort of persons who had a primary omicron infection, but different vaccination histories Dec 2021- Sept 2022 (= BA.1/2 and BA.4/5 – BA.2.75 period) in Qatar
Vaccination (as compared to unvaccinated) protected against re-infection:
- Hazard Ratio (HR) of 2 doses: 0.43 (CI 0.38-0.48)
- HR of 3 doses: 0.53 (CI 0.44-0.63)
However: incidence of re-infection higher in 3 doses as compared to 2 doses HR 1.38 (CI 1.16-1.65)
Effect of subvariant: Divergence in cumulative incidence curves in all comparisons increased markedly when incidence was dominated by BA.4/BA.5 and BA.2.75*
Remark: No severe, critical or fatal COVID in any cohort.
Primary (2 dose-series vaccination) followed by a primary omicron infection enhanced immune protection against omicron reinfection.
However, booster vaccination followed by a primary omicron infection compromised protection against omicron reinfection, perhaps because it involved repeat pre-omicron immunological events before the omicron infection.
These findings do not undermine the utility of booster vaccination in the short-term, but may point to potentially significant public health complexities requiring fine-tuning of booster vaccination to those who can best benefit from it, such as those most clinically vulnerable to severe COVID-19.
Ep 332-3: Nabin Shrestha medRxiv March 2023: Effectiveness of Bivalent (WT-BA.4/5) vaccine in US HCW
Retrospective cohort of documented omicron infections in Cleveland Clinic employees from Sept 2022 till med Jan 2023 (BA.4/5 – BQ – XBB variants).
- Protection against infection after bivalent booster decreased across variants
- BA.4/5: HR = 0.71 (0.63-0.79) = modest
- BQ : HR = 0.80 (0.69-0.94) = little
- XBB : 0.96 (0.80-1.12) = NO protection
- Risk of re-infection increased with time since previous infection (as expected)
- Risk of infection also increased with increasing doses of vaccine (= unexpected)
Fig 1: Recent (= omicron) infection protects better against re-infection than pre-omicron infection
Fig 2: More infections with increasing number of vaccines !
- Most positive tests would have been done to evaluate suspicious symptoms. Some would have been done to evaluate known exposures or tests for pre-operative or pre-procedural screening.
- Too few severe illnesses for the study to be able to determine if the vaccine decreased severity of illness.
- Healthcare population included no children and few elderly subjects, the majority would not have been immunocompromised.
- Overall modest protective effect of the bivalent vaccine against COVID-19 while the circulating strains were represented in the vaccine (= BA.4/5 period) and lower protection when the circulating strains were no longer represented.
- The unexpected finding of increasing risk with increasing number of prior COVID-19 vaccine doses needs further study.
Discussion: In these 3 papers, we see a counter-intuitive effect that additional vaccination could increase the likelihood of infection (but without firm data on disease severity). Especially the third paper by Shrestha is intriguing, because there is a paradox:
- there is a very modest overall protective effect of the bivalent vaccine against BA.4/5 and BQ (but not XBB),
- But there is also a dose-dependent increase of infection across all subvariants (Fig 2).
Interpretation of colleague Geert Molenberghs:
We should think in the direction of 'confounding' or Simpson's paradox. (1, 2)
First, let me give another example of Simpson's paradox. There are several times when the vaccine appears to dramatically increase the likelihood of hospitalization. However, when corrected for age, we see a strong protective effect within each stratum of age. This is textbook confounding: the elderly have a much higher chance of ending up in the hospital than the young, and are also more vaccinated (the two are, of course, related). Due to the enormous risk differences with age (a vaccinated elderly person has a higher risk of HD admission than an unvaccinated young person), the vaccine appears to entail a risk of hospitalisation. But in each stratum of age you see that the chance of HD uptake decreases.
Back to the problem, which is different, but there is a relationship. Only I am thinking less about age (which is already taken into account) but about history of infection. Someone who is less vaccinated may have had more infections, and gets the protection from that. The complicated combination of infection + vaccination history (where there are really many different possibilities, and if one takes event history into account, it is almost unique) means that only looking at vaccination history is probably insufficient. You therefore need data from a sufficiently representative sample for which the vaccination and infection history are also very well available. For the sake of asymptomatic or unreported mildly symptomatic infections, it is virtually impossible to have that fully covered.
Of course, a completely different explanation is that there is a certain priming for further (mild) infection, but protection against the more serious forms.
- Simpson’s Paradox is a statistical phenomenon where an association between two variables in a population emerges, disappears or reverses when the population is divided into subpopulations. For instance, two variables may be positively associated in a population, but be independent or even negatively associated in all subpopulations. See https://plato.stanford.edu/entries/paradox-simpson/
- Simpsons paradox occurs when the marginal association between two categorical variables is qualitatively different from the partial association between the same two variables after controlling for one or more other variables. Simpson’s paradox is important for three critical reasons.
- First, people often expect statistical relationships to be immutable. They often are not. The relationship between two variables might increase, decrease, or even change direction depending on the set of variables being controlled.
- Second, Simpson’s paradox is not simply an obscure phenomenon of interest only to a small group of statisticians. Simpson’s paradox is actually one of a large class of association paradoxes.
- Third, Simpson’s paradox reminds researchers that causal inferences, particularly in nonexperimental studies, can be hazardous. Uncontrolled and even unobserved variables that would eliminate or reverse the association observed between two variables might exist.
Par 2 Reports on favorable effects
Note; most of the following studies are “test-negative case-control” = involve comparing the odds of a given intervention (e.g., vaccine receipt) among symptomatic persons who test positive compared to those (matched symptomatics) who test negative.
Many adjustments need to be made for matching e.g.: age, gender, race, ethnicity, social vulnerability index and region of the testing location, underlying conditions (presence versus absence), local incidence (cases per 100,000 by individual county and state in the 7 days before test date) and date of testing.
Ep 332-4: Ruth Link-Gelles CDC 19 April VE of monovalent and BA.4/5 bivalent vaccine in children and adults US
- Children 3-5 years old: monovalent vaccine against symptomatic disease
Clear-cut, but not complete, protection by both Moderna and Pfizer monovalent vaccines against symptomatic infection during the omicron waves (BA.4/5, BQ and XBB)
- Adults protection against severe disease by bivalent booster during Omicron BA.4/5 and XBB period (Sept 2022- March 2023)
- Protection against hospitalization
Weak protection by monovalent doses only (> 2 months earlier)
Strong protection by bivalent vaccines especially during first 2 months in immunocompetent adults (both younger and older) see left graphs
Weaker, but still significant protection by bivalent vaccines in immunocompromised adults
- Protection against mechanical ventilation and death is even stronger
The complete meeting can be viewed at https://www.cdc.gov/vaccines/acip/meetings/slides-2023-04-19.html
Ep 332-5: Grewal medRxiv 11 April: VE of monovalent and bivalent mRNA against severe disease in adults over 50 in Ontario between 19 June 2022 and 28 Jan 2023.
Apparently very high VE against severe disease (= hospitalization or death) for both Pfizer and Moderna, irrespective whether monovalent, BA.1 bivalent or BA.4/5 bivalent.
Some waning observed during BQ period.
Ep 332-6: Eero Poukka medRxiv March: VE of bivalent booster against severe disease in elderly and “chronically ill adults” in Finland Sept 2022-Jan 2023
This is not a test-negative study, but a population-based cohort in over 1 million elderly (65 – 120 years old) and in over 400,000 “chronically ill” (co-morbidities or medical therapies predisposing to severe COVID) adults (18-64).
VE against hospitalization or death after bivalent booster (either BA.1 or BA.4/5) is compared with 2 doses (while in the previous studies unvaccinated people were the reference).
The outcome is different in both groups:
- Clear-cut, but waning protective effect in elderly over 65, and even in those over 80.
- No significant effect in chronically ill adults under 65.
Ep 332-7: Shangchen Song medRxiv Feb 2023: A Systematic Review and Meta-Analysis of 42 Test-Negative Design Studies on VE against Omicron.
Overall summary of VE within 60 days
Protection against infection waned after 60 days, but not so for protection against severe disease
- Pure mRNA vaccines provided comparable protection to partial mRNA vaccines, but both provided higher protection than non-mRNA vaccination.
- Also heterogeneity between studies with same vaccine
Protection against infection by 2 doses after first booster
Protection against severe disease by 2 doses after first booster
Par 3 Immunological aspects
Ep 332-8: Pascal Irrgang Sci Immunol Jan 2023: Class-switch towards non-inflammatory IgG4 isotype may explain decreased efficacy of third mRNA vaccine
Comparison of functional antibody responses after two or three mRNA vaccinations
sVNT (WT) = surrogate virus neutralization test against wild type; pVNT = psuodovirus neutralization against Omicron
ADCP = antibody-dependent cellular phagocytosis; ADCD = antibody-dependent complement deposition
A and B: a third mRNA vaccination increases the total anti-Spike IgG, but this is mainly due to a dramatic increase in IgG4, mainly oriented against the receptor binding domain (RBD).
C, E, F: third mRNA results in increased affinity (avidity) and also increased neutralization of wild-type and omicron.
HOWEVER: G, H, I: antibody-dependent cellular phagocytosis and complement deposition are reduced, because IgG4, in contrast to IgG1 and 2 fails to activate phagocytes (via Fc receptor) and complement.
Discussion: It is unclear at present which of these in vitro test is the best correlate of protection, but, if antibody functions such as stimulation of phagocytosis and complement activation are critical for control of SARS-CoV-2, it might explain why multiple mRNA doses have a decreasing effect on protection.
Ep 332-9: Yisimayi bioRxv 3 May Repeated Omicron vaccination or infection alleviates SARS-CoV-2 immune imprinting
This is a very complex and comprehensive paper from Yunlong Cao’s group in Bejing, who pioneered the research on immune imprinting. The messages are
As we know: Immune imprinting induced by WT (wild type)-based vaccination compromises the antibody response to Omicron-based boosters, which could result in reduced protection against successive Omicron subvariants.
- In mice, the efficacy of single Omicron-boosting is heavily limited by immune imprinting, especially when using variants antigenically distinct from WT, like XBB
BUT this concerning situation can be overcome by a second Omicron booster.
a, Comparison of neutralizing antibody responses among mouse immunization groups against pseudotyped D614G and additional SARS-CoV-2 variants. Mice sera antibody titers after priming with two doses of 3 μg CoronaVac followed by one boosting with 10 μg SARS-CoV-2 variant Spike proteins. Four different variants include BA.1, BA.5, BQ.1.1 and XBB
d, Neutralizing antibody response after priming with 2 doses of 3 μg CoronaVac followed by one boosting twice with 10 μg SARS-CoV-2 variant Spike proteins with different time intervals in mice. Mice were boosted twice with intervals between boosters of either 1 month or 3 months.
f, Neutralizing antibody response after priming with 2 doses of 3 μg CoronaVac followed by boosting twice with 1 μg or 10 μg of SARS-CoV-2 variant Spike mRNAs. 10 mice were immunized and analyzed in each group (n= 10). The time interval between the two priming doses is 2 weeks.
All Spike protein and mRNA used S6P prefusion-stabilization.
Fold changes between groups receiving different boosters and the primary-only group were calculated and shown above the line. Statistical significance was determined using the Wilcoxon rank sum test.
- A single boost after basic CoronaVac (= inactivated Chinese vaccine) with variant specific protein induces lower neutralizing Ab against the corresponding variant than against WT (D614G) = imprinting
(d and f) Boosting twice with either variant protein (d) or mRNA (f) results in equivalent neutralization titers against the variant than against WT.
- In humans, repeated Omicron infections could also alleviate WT-vaccination-induced immune imprinting and generate high neutralizing titers against XBB.1.5 and XBB.1.16
Two different omicron infections result in similar neutralization against a range of Omicron variants, compaerable to WT (D614G).
- Analysis of 781 RBD-specific monoclonal antibodies from double Omicron exposure
- Omicron-specific antibodies target distinct RBD epitopes compared to WT-induced antibodies
- The bias towards non-neutralizing epitopes observed in single Omicron exposures due to imprinting was largely restored after repeated Omicron stimulation.
- WT component should be abandoned when updating COVID-19 vaccine compositions to XBB lineages,
- People who haven't been exposed to Omicron yet should receive two updated vaccine boosters.
- Booster vaccination with either monovalent of bivalent mRNA has a limited protective effect on infection with subsequent omicron subvariants. The effect is heterogeneous amongst studies from weak to moderate (Ep 332-7). Some even suggest an enhancement (Ep 332-1, -2, -3), but those were retrospective studies and may have been complicated by Simpson’s paradox?
- Booster vaccination remains very protective against severe disease, also in elderly. Immune compromised subject, however, enjoy less protection.
Imprinting may be overcome by TWO doses of updated MONOVALENT variant vaccines (without WT component).
If you have additional information or insights, please let me know and I will share it with the other colleagues.
It is important for us as “experts” to share a consensus view on how to deal with SARS-CoV-2 vaccination in order to communicate unambiguously with the general public.
7 May 2023 Episode 332 Critical evaluation on COVID vaccine effectiveness during successive omicron waves
> More info
4 April 2023 Episode 326 Adeno-associated viruses as a cause of hepatitis and as a tool for gene therapy
> More info