Age-related Imaging, Inflammatory, and Immune Phenotypes of Varicella Pneumonia Enabled by AI-derived CT Quantification
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background Varicella pneumonia is one of the major complications of varicella-zoster virus infection; however, age-related differences in imaging burden and inflammatory–immune responses remain insufficiently characterized. Artificial intelligence (AI) – driven Computed tomography (CT) quantification offers an objective approach for evaluating lesion burden and disease heterogeneity. Objectives This study aimed to characterize the age-specific imaging, inflammatory, and peripheral immune profiles of varicella pneumonia using AI-derived CT quantitative biomarkers—the Nodular Load Index (NLI) and whole-lung infection ratio—and to explore their potential roles in risk stratification and clinical management. Methods In this dual-center retrospective study, 104 patients with varicella pneumonia were categorized into children (< 10 years, n = 28), adolescents (10–19 years, n = 41), and adults (> 19 years, n = 35). An AI-CT pipeline automatically extracted nodule counts, total nodule volume, and mean nodule attenuation; these were standardized and summed to construct the NLI. Whole-lung infection ratio was quantified simultaneously. Clinical variables, laboratory parameters, and CT-derived metrics were compared across age groups. Spearman correlation assessed associations between imaging biomarkers and systemic inflammatory and peripheral T-cell immune indicators. Results Adults exhibited significantly higher NLI and C-reactive protein (CRP) levels than children and adolescents (all P < 0.05), indicating greater nodular inflammatory burden and systemic inflammation. Children showed higher white blood cell counts (WBC) and absolute CD4⁺ T-cell numbers compared with adults (P < 0.05), reflecting more active peripheral immune responses. NLI was negatively correlated with neutrophil percentage, procalcitonin, and CD4⁺/CD8⁺ T-cell ratio, but positively correlated with lymphocyte percentage and the proportion of CD8⁺ T cells among CD3⁺ T cells (all P < 0.05). No significant associations were observed between whole-lung infection ratio and inflammatory or immune parameters. Conclusions Varicella pneumonia demonstrates marked age-related heterogeneity in imaging, systemic inflammation, and peripheral immune responses. Adults present with higher NLI and CRP, whereas children show more active peripheral immune responses. AI-derived NLI accurately quantifies hematogenously disseminated “nodule-plus-halo” lesions and correlates with peripheral immune status, providing a promising biomarker for disease stratification and clinical management. Prospective studies are warranted to validate its robustness and prognostic value.