COVID Vaccine Harms: What the 2025 JPandS Review Argues

In late 2025 a group of eleven physicians and researchers published a review in the Journal of American Physicians and Surgeons arguing that COVID-19 modified mRNA products have caused widespread systemic harm. It's been shared heavily. This article explains what the paper actually claims, where each claim comes from, and which parts hold up to a look at the underlying sources.

The point isn't to endorse or dismiss it. It's to show you how to read it.

What the paper is

The review is COVID-19 Injections: Harms and Damages, a Non-Exhaustive Conclusion, by Andrew Zywiec, Irene Mavrakakis, Peter McCullough, Nicolas Hulscher, Aaron Kheriaty, Paul Marik, James Thorp, Marivic Villa, Charles Rixey, Edward Macie and Abraxas Hudson, published in volume 30, number 3, pages 80 to 89.

Two things are worth knowing up front. First, it's a narrative review: it assembles and interprets other people's findings rather than reporting new data. Second, JPandS is published by the Association of American Physicians and Surgeons and is not indexed in MEDLINE. The venue doesn't decide whether a claim is right, but it does mean the cited sources have to do all the work.

Next step

Want complex vaccine research explained without the noise?

Explore our library of plain-English guides on spike protein, microclots, and post-viral recovery.

See the at-home test

The mechanistic claims

Ribosomal frameshifting

This is the review's strongest citation. To keep vaccine mRNA stable, manufacturers replaced uridine with N1-methylpseudouridine. In 2024, Mulroney and colleagues demonstrated in Nature that this modification causes +1 ribosomal frameshifting, and that cellular immunity to the frameshifted products occurs in mice and humans after BNT162b2 vaccination (Mulroney et al., 2024).

Picture a sentence where every word is three letters: THE CAT ATE THE RAT. Slip one letter and you read T HEC ATA TET HER AT. The output stops being the intended message.

What the Nature authors say next matters and is usually dropped in summaries: they state that no adverse outcomes from mistranslation of mRNA SARS-CoV-2 vaccines have been reported in humans, and they show that redesigning the slippery sequences reduces frameshifted products. The finding is real. The leap from "unintended proteins are produced" to "this is causing the harms described" is the authors' argument, not Mulroney's finding.

Spike persistence

The review argues that spike protein persists far longer than the original pharmacokinetic picture implied. This part has the most independent support. Patterson and colleagues reported S1 in CD16+ monocytes up to 15 months after infection (Patterson et al., 2022), and several groups have detected vaccine-derived material or spike antigen weeks to months after dosing. Our article on spike antibody versus antigen testing walks through what each of those studies measured and what detection does and doesn't prove.

Persistence of an antigen is not the same as persistence of harm. It's a plausible starting point for mechanism, which is why the spike amyloid and fibrinolysis research matters: it proposes a specific thing persistent spike fragments might do. Parallel research into lipid nanoparticle membrane dysfunction also examines how carrier components cause cellular stress independently.

Immune dysregulation

The review cites a systematic review by Rodríguez and colleagues covering 928 post-vaccination autoimmune cases, of which 81.5% were new onset, with immune thrombocytopenia, myocarditis and Guillain-Barré syndrome the most common new-onset conditions (Rodríguez et al., 2022). That paper is a case-based literature review: it describes reported cases, not incidence, and it can't establish how often these events occur per million doses.

It also cites IgG4 class switching after repeated boosters (Uversky et al., 2023), itself a hypothesis-driven paper proposing that the shift may generate immune tolerance.

A Florida cohort of 817 vaccine-injured retirees supplies several striking percentages in the review, including autoimmune markers in roughly 75% of patients. Checking the reference list, those figures are attributed to unpublished preliminary observational outpatient clinic data from a single private clinic. Unpublished single-clinic data in a self-selected patient group cannot support population-level conclusions, and the review doesn't frame it that way.

The epidemiological claims

What the large cohorts found

The Global Vaccine Data Network study of 99 million vaccinated people is real and widely cited. It identified elevated observed-to-expected ratios for myocarditis and pericarditis after mRNA vaccines, and for Guillain-Barré syndrome and cerebral venous sinus thrombosis after viral vector vaccines (Faksova et al., 2024). Those myocarditis and pericarditis signals are not contested; they appear in regulatory safety reporting worldwide. Whole-body imaging of lipid nanoparticle biodistribution later demonstrated in mice that mRNA-carrying nanoparticles reach heart capillary walls, providing a potential biological mechanism for localized cardiac inflammation.

Where a citation doesn't match its source

The review attributes to a Bayesian meta-analysis a 286% increased risk of myocardial infarction after dose two and a 240% increased risk of stroke after dose one. The published paper reports something different. Karimi and colleagues found an association with coronary artery disease risk, and explicitly state that no increased risk of heart attack, arrhythmia or stroke was observed after COVID-19 vaccination, with a protective effect on stroke and myocardial infarction after the third dose (Karimi et al., 2025).

That's a substantial mismatch between a review's summary and its source. It's also the single best argument for reading the primary papers rather than the review.

The autopsy series

The review leans on an analysis of 325 autopsy cases as pathological confirmation of vaccine-induced mortality. The version of that systematic review published in Forensic Science International in 2024 was withdrawn, and an earlier preprint posting was also retracted. Whatever one makes of the underlying case material, a retracted source can't function as confirmation of anything, and the review doesn't tell readers it was retracted.

Military database figures

Defense Medical Epidemiology Database comparisons are presented as showing large elevations in myocarditis, pulmonary embolism, clotting defects and certain cancers. DMED counts medical billing codes. Coding practice, care-seeking behaviour, deployment patterns and the choice of baseline years all shift those numbers, and the Department of Defense has disputed earlier analyses on data-quality grounds. Treat these as a signal that warrants investigation, not as a measurement of incidence.

The cancer argument

The review assembles a multi-hit model: T-cell exhaustion, IgG4 class switching, spike S2 interference with p53, and residual plasmid DNA.

The p53 element is worth looking at closely, because it's frequently overstated. Zhang and El-Deiry examined spike, p53 and MDM2 in cancer cells and found that spike interrupts the p53-MDM2 interaction and suppresses p53 transcriptional activity, but they explicitly did not detect spike bound to p53 in cells (Zhang & El-Deiry, 2024). It's a cell-culture study of transfected spike, offering a potential molecular mechanism. It is not evidence that vaccination causes cancer in people.

The "turbo cancer" clinical pattern the review describes rests on clinician observation rather than registry data. Cancer registries take years to settle questions of incidence, and that work hasn't been done.

Reproductive claims

The review summarises re-analyses of the Shimabukuro pregnancy data and figures from Pfizer's post-marketing documents. These re-analyses are contested in the literature, and the denominators and follow-up windows are exactly where the disagreement lives. We're not going to reproduce percentages that depend on contested denominators.

One cited finding is straightforwardly published: Lin and colleagues reported evidence of vaccine mRNA in placental and cord blood samples (Lin et al., 2024). Detection of mRNA is a measurement. What it implies clinically is a separate question that study doesn't answer.

How to grade the review's claims

What's actually useful here for patients

Strip away the framing and a few things remain worth taking seriously. Spike protein persistence is documented. Spike-derived amyloid fibrils can change how fibrin clots behave, at least in purified systems, which our walkthrough of how neutrophil elastase generates those fragments explains from the start. Amyloid-positive microclot material has been reported in both Long COVID cohorts and a small post-vaccination cohort. People with persistent symptoms after infection or vaccination deserve objective workup rather than dismissal.

None of that requires accepting the review's broader conclusions. It just means the underlying biology is a legitimate research area, which is also why the field is actively developing alternatives to lipid nanoparticle delivery.

What to do with symptoms

If your routine labs came back normal and your symptoms haven't, the productive route is more measurement, not more argument. That means inflammatory markers, coagulation markers, autoantibody screening, and a documented timeline of exposures and symptom onset. Our comparison of D-dimer versus microclot testing explains why standard clotting panels can look fine while amyloid-positive microclots are present, and the free symptom screener is a starting point for organising what you're experiencing.

Our pillar guide to amyloid fibrin testing explains what that kind of microscopy can and can't establish. Our at-home amyloid test contributes one objective observation to the picture: what Thioflavin T fluorescence microscopy sees in a dried blood sample. It's laboratory observational data for informational and research purposes, not a diagnosis, not a medical device, and not a measure of spike protein. Take any result to a clinician who can read it against your full history, and if you need one, the doctor directory lists practitioners working in this area.