Can we estimate cytokine storm from initial computed tomography images of Coronavirus disease-2019 patients?

OBJECTIVES: The present study aims to investigate whether elementary lesions detected at the time of the diagnosis, their distribution characteristics, and CT scoring can be predictive of a cytokine storm. BACKGROUND: CT might have a prognostic predictive value beyond its diagnostic value. METHODS: Sixty-eight patients, 32 with cytokine storm and 36 without cytokine storm, were included in the study. Four different scoring methods were created according to elementary lesions, distribution and involvement rate. CT scores and demographic findings of the cases were compared in the cytokine storm and non-cytokine storm groups. RESULTS: The mean age of patients was 57.72 (SD: 13.5) and 40 (58.8 %) of them were male. The cytokine storm was significantly more common among male patients and patients of older age (p=0.04).  The AUC values of CT score 1, CT score 2, CT score 3, and CT score 4 were as follows; 0.772 (95% CI; 0.651-0.892), 0.766 (95% CI; 0.647-0.885), 0.758 (95% CI; 0.639-8.78), and 0.760 (95% CI; 0.640-0.881), respectively. All CT scores had better predictive values in males. CONCLUSIONS: CT scoring at the time of admission can be used to predict cases that may develop cytokine storm later (Tab. 4, Fig. 2, Ref. 15).