Composite Radiological Score (CRS)-A novel imaging biomarker: Correlation with clinical outcomes in patients with COVID-19 pneumonia

Abstract

Background

Computed Tomography (CT) was widely used for COVID-19 patients to assess extent of lung abnormalities and disease severity. Most CT scoring systems have relied on the extent of lung involvement.

Objectives

We developed a novel CT scoring system- Composite Radiological Score (CRS) incorporating both extent and density of lung lesions to assess the trends in CRS at admission, 4–8 weeks and 10–12 weeks post-admission, estimate inter-reader agreement and correlate CRS with clinical severity.

Methods

A prospective, observational study of consecutive adult patients admitted with RT-PCR confirmed COVID-19 pneumonia (June to September 2020) in a tertiary hospital in North India. Clinical data, CT images and CRS were recorded sequentially at admission (CT1) and 4–8 weeks (CT2) in all patients, and at 10–12 weeks (CT3) in a subgroup of patients.

Results

CRS was significantly associated with clinical severity (p-value = 0.003), d-dimer (p-value = 0.004), serum ferritin (p-value = 0.03) and median length of stay (p-value = 0.02). At 4–8 weeks, proportion of patients with CRS>3 (15 %) was significantly lower than those with the extent of lung involvement>25 % (30 %). Kappa agreement with regards to density of lesions and CRS score was moderate to substantial except for one pair of radiologist where fair agreement was recorded. Higher kappa agreement was seen in CT2 and CT3 compared to CT1.

Conclusion

CRS correlated well with clinical severity and other laboratory parameters of disease severity. CRS provides a new method to non-invasively assess the severity of diffuse lung abnormalities of COVID-19 pneumonia that incorporates both the extent and density of lung lesions.

Clinical relevance

CRS will be useful to assess severity and longitudinally monitor clinical course of patients with airspace predominant lung abnormalities.