Abstract

Background

Authorized COVID-19 vaccine platforms encode or deliver spike protein via distinct delivery architectures and excipient compositions. Among these, mRNA platforms use ionizable lipid nanoparticles (LNPs) with PEG-lipid excipients, while adenoviral vector platforms use free polysorbate 80 (PS80) as a stabilizing surfactant. Whether the tissue-specific adverse event profiles observed across platforms reflect antigenic or excipient differences has not been systematically evaluated.

Methods

We performed a comparative pharmacovigilance analysis combining literature-derived disproportionality estimates with original signal detection in VAERS OpenData (all reports through 29 May 2026; n = 1,924,924). Reporting Odds Ratios (ROR) with 95% confidence intervals were calculated for myocarditis/pericarditis/myopericarditis and for a combined POI/POTS/amenorrhoea endpoint across three COVID-19 vaccine platforms (Janssen, Moderna, Pfizer–BioNTech) and one free PS80 reference comparator (Gardasil). NVX-CoV2373 was excluded from the primary analysis because its Matrix-M adjuvant introduces independent membranolytic activity that prevents attribution of any signal to the PS80 micellar component alone. Platforms were classified by excipient architecture: free PS80 adenoviral and ionizable PEG-lipid LNP.

Results

Myocarditis disproportionality followed excipient architecture rather than antigenic content: PEG-lipid LNP platforms showed significant signal (Pfizer ROR 4.16, 95% CI 3.95–4.38; Moderna ROR 1.51, 95% CI 1.43–1.59), while the free PS80 adenoviral platform showed no signal (Janssen ROR 1.01, 95% CI 0.88–1.14), despite all three platforms encoding identical spike protein. The POI/POTS signal showed the inverse pattern: enriched in free PS80 platforms (Janssen ROR 1.42; Gardasil ROR 12.68) relative to PEG-lipid LNP platforms (Pfizer ROR 2.57; Moderna ROR 1.10). The POI/POTS:myocarditis ratio ranged from 0.62–0.73× in PEG-lipid LNP platforms to 1.41× in the free PS80 adenoviral platform and 71.47× in Gardasil.

Interpretation

We propose the Fusogenic Architecture Hypothesis: excipient supramolecular architecture — specifically, the capacity for endosomal membrane fusion — may be a major determinant of tissue-specific pharmacovigilance signals. Ionizable LNPs drive cardiomyocyte damage through endosomal rupture; free PS80 drives ovarian and autonomic damage through ApoE-mediated transcytosis into immunologically privileged compartments. This double dissociation is incompatible with spike-antigen-centered explanations and points to excipient architecture as an independent pharmacological variable requiring systematic evaluation in vaccine safety frameworks. Five falsifiable predictions are provided.