Chest complications in immunocompromised patients without acquired immunodeficiency syndrome (AIDS): differentiation between infectious and non-infectious diseases using high-resolution CT findings

N Tanaka,Y Kunihiro,R Kawano,T Yujiri,K Ueda,T Gondo,T Matsumoto
DOI: https://doi.org/10.1016/j.crad.2020.07.029
Abstract:Aim: To differentiate between infectious and non-infectious diseases occurring in immunocompromised patients without acquired immunodeficiency syndrome (AIDS) using high-resolution computed tomography (HRCT). Materials and methods: HRCT images of 555 patients with chest complications were reviewed retrospectively. Infectious diseases (n=341) included bacterial pneumonia (n=123), fungal infection (n=80), septic emboli (n=11), tuberculosis (n=15), pneumocystis pneumonia (n=101), and cytomegalovirus pneumonia (n=11), while non-infectious diseases (n=214) included drug toxicity (n=84), infiltration of underlying diseases (n=83), idiopathic pneumonia syndrome (n=34), diffuse alveolar haemorrhage (n=8), and pulmonary oedema (n=5). Lung parenchymal abnormalities were compared between the two groups using the χ2 test and multiple logistic regression analysis. Results: The χ2 test results showed significant differences in many HRCT findings between the two groups. Multiple logistic regression analysis results indicated the presence of nodules with a halo and the absence of interlobular septal (ILS) thickening were the significant indicators that could differentiate infectious from non-infectious diseases. ILS thickening was generally less frequent among most infectious diseases and more frequent among most non-infectious diseases, with a good odds ratio (7.887, p<0.001). The sensitivity and accuracy for infectious diseases in the absence of ILS thickening were better (70% and 73%, respectively) than those of nodules with a halo (19% and 48%, respectively), while the specificity in the nodules with a halo was better (93%) than that of ILS thickening (78%). Conclusions: The presence of nodules with a halo or the absence of ILS thickening tends to suggest infectious disease. Specifically, ILS thickening seems to be a more reliable indicator.
What problem does this paper attempt to address?