Suppnonsense-mediated decay-linked mutations in SARS-CoV-2 and their association with COVID-19 disease severity

Nonsense-mediated decay (NMD) is a cellular mechanism that degrades mRNAs with premature termination codons (PTCs), preventing the production of truncated, potentially harmful proteins. While its role in viral infections is increasingly recognized, the relationship between NMD-linked mutations in SARS-CoV-2 and COVID-19 severity remains poorly understood. To investigate the presence of SARS-CoV-2 nonsense mutations predicted to trigger NMD and assess their association with clinical disease severity and viral genomic. We conducted whole-genome sequencing on samples from 129 hospitalized COVID-19 patients. A panel of 21 nonsense mutations predicted to activate the NMD pathway was identified and analyzed. Statistical correlations with clinical severity were assessed, including multivariate analysis. Receiver Operating Characteristic (ROC) curves and interdependency analysis of mutation combinations were also performed. Five NMD-associated mutations (Variants 5, 6, 7, 9, and 15) showed significant associations with mild disease. These mutations, located in key SARS-CoV-2 genes, include Variant 5 (g.C5575G, T, synonymous substitution in ORF1ab), Variant 6 (g.T5653G, substitution in ORF1ab), Variant 7 (g.G6094GGACAGACTTT/GCCTACACGACGCTAATC, insertion in spike protein), Variant 9 (g.G6446GAATGA, insertion in spike protein), and Variant 15 (g.T10968TATATTGA, insertion in N protein). In the host-adjusted multivariable model, the presence of at least one NMD-inducing mutation was an independent protective factor (OR = 0.34, 95% CI: 0.16–0.72, p = 0.005). However, after adjusting for SARS-CoV-2 lineage, this association was attenuated (OR = 0.67, 95% CI: 0.03–13.84, p = 0.793). In the lineage-only model, Omicron infection showed higher odds of severe disease compared to Delta (OR = 1.91, 95% CI: 0.77–4.77, p = 0.164). ROC analysis indicated limited predictive value for individual variants (AUC = 0.37–0.48), but specific combinations, such as Variants 5 and 7, markedly reduced severe case incidence (92.9–4.8%, p = 0.0001). NMD-inducing nonsense mutations were associated with reduced COVID-19 severity in host-adjusted analyses, but this effect diminished after accounting for viral lineage, suggesting that variant distribution, particularly Omicron may influence these associations. Integrating viral genomic background with host and clinical data may enhance risk prediction and inform antiviral strategies.