Relationship between CT air trapping criteria and lung function in small airway impairment quantification

Abstract

Background: Small airways are regarded as the elective anatomic site of obstruction in most chronic airway diseases. Expiratory computed tomography (CT) is increasingly used to assess obstruction at this level but there is no consensus regarding the best quantification method. We aimed to evaluate software-assisted CT quantification of air trapping for assessing small airway obstruction and determine which CT criteria better predict small airway obstruction on single breath nitrogen test (SBNT).

Methods: Eighty-nine healthy volunteers age from 60 to 90 years old, underwent spirometrically-gated inspiratory (I) and expiratory (E) CT and pulmonary function tests (PFTs) using SBNT, performed on the same day. Air trapping was estimated using dedicated software measuring on inspiratory and expiratory CT low attenuation area (LAA) lung proportion and mean lung density (MLD). CT indexes were compared to SBNT results using the Spearman correlation coefficient and hierarchical dendrogram analysis. In addition, receiver operating characteristic (ROC) curve analysis was performed to determine the optimal CT air-trapping criterion.

Results: 43 of 89 subjects (48,3%) had dN2 value above the threshold defining small airway obstruction (i.e. 2.5% N2/l). Expiratory to inspiratory MLD ratio (r = 0.40) and LAA for the range -850 -1024 HU (r = 0.29) and for the range -850 -910 HU (r = 0.37) were positively correlated with SBNT results. E/I MLD was the most suitable criterion for its expression. Expiratory to inspiratory MLD ratio (E/I MLD) showed the highest AUC value (0.733) for small airway obstruction assessment.

Conclusion: Among all CT criteria, all correlating with small airway obstruction on SBNT, E/I MLD was the most suitable criterion for its expression in asymptomatic subjects with mild small airway obstruction

Trial registration: Registered at Clinicaltrials.gov, identifier: NCT01230879.

Figures

Figure 1

Box plots of air trapping represented by E/IMLDin 4 levels and mean of E/IMLD. Significant differences between slices were observed (p = 0.0002).

Figure 2

Hierarchical dendrogram analysis to evaluate distance and similarity between dN2 and CT air trapping criteria.

Figure 3

Receiver-operating characteristic (ROC) curves show the diagnostic performance of E/I using dN2 as a functional test of airway obstruction (dN2 cut-off = 2.5% N2/l). The best threshold is reported, alongside the sensitivity and specificity: E/I MLD = 0.89 (0.70; 0.73) – E/I -850–1024 = 0.32 (0.63; 0.67) – E/I -850–910 = 0.29 (0.79; 0.61).

Figure 4

Receiver-operating characteristic (ROC) curves show the diagnostic performance of E-I using dN2 as a functional test of airway obstruction (dN2 cut-off = 2.5% N2/l). The best threshold is reported, alongside the sensitivity and specificity: E-I MLD = 90.01 (0.79; 0.63) – E-I -850–1024 = −34.24 (0.65; 0.65) – E-I -850–910 = −47.23 (0.81; 0.59).

Figure 5

Receiver-operating characteristic (ROC) curves show the diagnostic performance of (E-I)/I using dN2 as a functional test of airway obstruction (dN2 cut-off = 2.5% N2/l). The best threshold is reported, alongside the sensitivity and specificity: (E-I)/I MLD = −0.11 (0.70; 0.73) – (E-I)/I -850–1024 = −0.68 (0.63; 0.67) – (E-I)/I -850–910 = −0.70 (0.79; 0.61).