Vision problems are among the least-discussed consequences of long COVID, yet for affected individuals they can be profoundly disabling. A new finding from Nature Communications shifts the conversation: persistent ocular symptoms after mild COVID-19 are not psychosomatic or transient — they carry measurable biological signatures that standard eye exams entirely miss, and they can endure for up to three years.
This prospective cross-sectional study enrolled non-hospitalized COVID-19 survivors and compared those with persistent ocular symptoms (POS) against recovered individuals without eye complaints. The affected group showed a constellation of findings: near-vision disturbances, acquired strabismus, weakened autonomic pupillary reflexes, corneal nerve fiber degeneration, and chronic immune activation at the ocular surface involving dendritic cells and T cells. Tear film proteomics revealed a molecular fingerprint strikingly similar to that seen in severe acute COVID-19 cases, featuring upregulation of integrin ITGB6, neurofascin (NFASC), connective tissue growth factor (CTGF), tryptase TPSAB1, and mitochondrial creatine kinases CKMT1A/B. Pupillary dysfunction correlated specifically with elevated JUN, a transcription factor implicated in neuronal stress responses, while immune dysregulation tracked with ANGPTL2, SKAP2, and DAPP1. Diagnostic models combining clinical and biomarker data predicted POS membership with 77–91% accuracy.
This work is significant for several reasons. First, it establishes that the eye is not a passive bystander in long COVID but an active site of chronic T cell-mediated neuroinflammation — a mechanism consistent with emerging data on long COVID's neuropathological footprint elsewhere in the nervous system. Second, the corneal neurodegeneration finding aligns with earlier smaller studies suggesting that confocal corneal microscopy may serve as a window into systemic small-fiber neuropathy, a known long COVID feature. The cross-sectional design limits causal inference, and the cohort is presumably self-selected toward symptomatic individuals, which may inflate effect sizes. Replication in larger, longitudinal samples is essential. Even so, the proteomic precision and diagnostic accuracy reported here suggest this syndrome is real, mechanistically grounded, and potentially tractable — making it a priority for both clinical recognition and therapeutic investigation.