Air trapping in usual interstitial pneumonia pattern at CT: prevalence and prognosis

This study was conducted to evaluate the presence of air trapping in patients with idiopathic pulmonary fibrosis (IPF) and other interstitial lung diseases (ILDs) (non-IPF), showing the radiological pattern of usual interstitial pneumonia (UIP). Retrospectively, we included 69 consecutive patients showing the typical UIP pattern on computed tomography (CT), and 15 final diagnosis of IPF with CT pattern “inconsistent with UIP” due to extensive air trapping. Air trapping at CT was assessed qualitatively by visual analysis and quantitatively by automated-software. In the quantitative analysis, significant air trapping was defined as >6% of voxels with attenuation between −950 to −856 HU on expiratory CT (expiratory air trapping index [ATIexp]) or an expiratory to inspiratory (E/I) ratio of mean lung density >0.87. The sample comprised 51 (60.7%) cases of IPF and 33 (39.3%) cases of non-IPF ILD. Most patients did not have air trapping (E/I ratio ≤0.87, n = 53, [63.1%]; ATIexp ≤6%, n = 45, [53.6%]). Air trapping in the upper lobes was the only variable distinguishing IPF from non-IPF ILD (prevalence, 3.9% vs 33.3%, p < 0.001). In conclusion, air trapping is common in patients with ILDs showing a UIP pattern on CT, as determined by qualitative and quantitative evaluation, and should not be considered to be inconsistent with UIP. On subjective visual assessment, air trapping in the upper lobes was associated with a non-IPF diagnoses.

Hence, air trapping is a common finding in IPF, and current guidelines stating that this CT finding is inconsistent with the diagnosis of IPF are likely too restrictive.
The primary aim of this study was to evaluate CT finding of air trapping with the in patients with IPF and those with other ILDs. A secondary aim was to examine the utility of different quantitative measurements of air trapping as CT biomarkers for the risk of death or lung transplantation in patients with IPF.

Participants. With the approval of the institutional review board of Santa Casa de Misericórdia de Porto
Alegre Hospital and waiver of the requirement for patient consent, we retrospectively identified 101 consecutive patients who presented with the UIP pattern on chest CT between July 2012 and December 2017, regardless of the presence of air trapping on CT. Patients were identified in our weekly multidisciplinary team meetings, with discussion of all cases according to the 2011 ATS/ERS/JRS/ALAT guidelines 1 . Clinical data recorded in patients' medical records, such as possible exposures, presence of CTDs, autoimmune features, and follow-up data, were reviewed. All imaging exams and histological samples were revised. This study was conducted in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments.
Inclusion criteria were a UIP pattern at chest CT defined as the presence of all the following four features: subpleural and basal predominance; reticular abnormality; honeycombing with or without traction bronchiectasis; and absence of features listed as inconsistent with UIP pattern 1 . Additionally, we also included all patients (n = 15) showing a definite UIP pattern on inspiratory CT, and extensive (i.e., in three or more lobes) air trapping, despite this feature's characterization as inconsistent with UIP 1 . Patients were excluded due to lack of follow-up with a pulmonologist after the initial CT scan or inadequacy of medical recording (n = 7); and initial CT findings inconsistent with the UIP pattern due to extensive air trapping and at least one other incompatible feature (n = 10).
Cases showing the UIP pattern on CT in the absence of any identifiable cause of another ILDs were diagnosed as IPF 1 . Patients with typical UIP pattern and clinical features suggestive of a non-IPF etiology (e.g., history of antigen exposure, clinical findings of CTD) underwent SLB for definitive diagnosis. For these cases, final diagnoses was established by multidisciplinary team consensus based on all clinical, imaging, and pathology findings. Patients were divided into IPF and non-IPF groups according to final diagnoses.
CT protocols. All subjects underwent a paired inspiratory and expiratory chest CT with 16 × 1.25 mm collimation (LightSpeed 16 Slice Pro, General Electric Healthcare Technologies, Waukesha, WI, USA). We followed the optimal CT protocol for evaluation of ILDs recommended by the ATS/ERS/JRS/ALAT guidelines 1 . Scans were performed caudocranially using a helical acquisition. Images were reconstructed with a slice thickness and interval of 1.0 mm to achieve near-isotropic voxels. Images were acquired using the following parameters: tube voltage, 120 kV; tube current for inspiratory image, 200 mAs; tube current for expiratory images, 50 mAs; pitch, 1.375. A standard reconstruction kernel was used to achieve medium-smooth images. Reconstructions were obtained in the sagittal and coronal planes. A data matrix of 512 × 512 and a field of view of 35-45 cm were used. No CT dose modulation or intravenous contrast agent was used for this study.
Imaging analysis. All examinations were analyzed in an Advantage Workstation (Advantage Workstation 4.6, General Electric Healthcare Technologies, Waukesha, WI, USA) from one PACS system. Inspiratory and expiratory CT images were evaluated using software designed for the assessment of segmented images from the chest wall, mediastinum, diaphragm, and airways. Automated segmentation of the right and left lungs from the chest wall and mediastinum was performed. Two chest radiologists, each with more than 8 years of experience, independently assessed the CT images blinded to the clinical information to visually qualify the presence of lobar air trapping. Images were analyzed according to criteria defined in the ATS/ERS 2011 guidelines in the Fleischner Society's Glossary of Terms 1,10 . The total lung volume and attenuation of all voxels included in the lung segmentation were quantified. The expiratory air-trapping index (ATIexp) was calculated as the percentage of lung voxels with attenuation of −950 to −856HU on expiratory CT images, to exclude emphysema areas 11 . Mean lung density (MLD) histograms were created for expiratory and inspiratory acquisitions for each subject, and the expiration and inspiration (E/I) ratio of MLD was calculated 12,13 . Air trapping was defined as ATIexp > 6% or E/I ratio > 0.87. These cutoff values were determined in previous studies on COPD and asthma, and we hypothesized that they could also be applied in ILDs 13,14 . The presence and extent of air trapping were scored visually in consensus by two radiologists. In cases of discrepancy, a third radiologist with more than 15 years of experience was consulted. Air trapping was assessed according to the lobes involved, defined as sharply demarcated areas of lung tissue with lesser than normal increase in attenuation and lack of volume reduction after expiration in more than five lobules 10,15 . Statistical analysis. Data was presented as frequency and percentage or mean ± standard deviation (SD).
Shapiro-Wilk test was used to test for normality. Continuous variables were compared using the independent Student's t-test. We assessed the association between categorical variables with the Chi-Square test. Kaplan-Meier curves compared with log-rank tests were used for cumulative survival analysis. Survival analysis was performed using Cox proportional hazard regression models: (i) Events were defined as the time to death or unilateral pulmonary transplantation; (ii) censored data was used when the event did not occur at the end of the follow-up period. A p-value < 0.05 level was considered significant. Data were analyzed using Stata software, version 13 (StataCorp, College Station, TX, USA). . IPF was diagnosed in 51 (60.7%) cases, by SLB (n = 10) and by multidisciplinary team consensus based on clinical findings and presence of the UIP pattern on CT (n = 41). In the non-IPF group (n = 33, [39.3%]), diagnoses were CHP (n = 12), rheumatoid arthritis (n = 9), CTDs (n = 6), microscopic polyangiitis (n = 3), granulomatosis with polyangiitis (n = 1), pneumoconiosis (n = 1), and amiodarone pulmonary toxicity (n = 1).  Figure 1 demonstrates the quantitative assessment of areas of air trapping in a patient with IPF. Figure 2 illustrates a case of a patient that presented an inconsistent with UIP pattern at CT due to extensive air trapping, but who had a diagnosis of IPF based on a UIP pattern in histopathology and multidisciplinary team consensus.
Among patients with IPF, we analyzed differences in the cumulative survival between patients with significant air trapping defined by quantitative CT measurements (i.e., E/I-ratio > 0.87 and ATIexp > 6%). Although these patients had slightly worse event-free survival at 1, 3, and 5 years, these differences were not significant (p = 0.085 and 0.188) (Fig. 3).

Discussion
Air trapping was a common finding in patients with ILD, present in up to 45.1% of patients with IPF in this sample. Moreover, the degree of air trapping did not differ between patients with IPF and those with UIP of known etiology according to either index used in this study. However, the finding of air trapping in the superior lobe is rare present in patients with IPF, and thus may be highly suggestive of an alternative diagnosis when present in patients with a UIP pattern.
The differential clinical diagnosis of radiological UIP is challenging, as it requires distinction of several conditions, including IPF 1-4 . Imaging findings may be similar to those for non-specific interstitial pneumonia, CHP, or sarcoidosis, which may lead to initial difficulty of diagnosis, especially when they are the only parameters considered 16 . The most recent (2011) ATS/ERS guidelines emphasized the importance of multidisciplinary discussion among radiologists, pulmonologists and pathologists for the diagnosis of IPF, to increase interobserver agreement 1,17 . In this study, cases showing the typical UIP pattern on CT in the absence of any identifiable cause of another ILD were diagnosed as IPF to avoid an unnecessary performance of invasive procedure 18 . However, when CT demonstrated air trapping in more than three lobes, patients underwent SLB to establish the final diagnosis. The greater mortality rate in patients with IPF relative to those with other types of UIP supports the accuracy of diagnoses in our study population.
Small airway obstruction with subsequent air trapping is a common finding in several pulmonary conditions, such as obstructive diseases, bronchiectasis, and ILD 6,19 . Although no cutoff value clearly defining normality has been established, current recommended for the quantitative measurement of air trapping state that the is the E/I ratio is the best method 20 . In our study, we have used the cutoff of 0.87, which was found to be the optimal cutoff to depict air trapping in previous studies on COPD an asthma 20 . In general, the greater the ratio, the greater the air trapping 20 . Among the various forms of ILD, CHP is characterized by air trapping as a main tomographic finding 16 . In certain types of CHP, HRCT may demonstrate the pattern of UIP, associated with a reduction in pulmonary volume, and the presence of linear and reticular opacities and honeycombing 21 . The presence of centrilobular nodules, lobular areas with decrease attenuation, and predominance in the superior regions of lung tend to differentiate CHP from IPF 8,22 . In our cohort, 11 of the 13 patients with air trapping in the superior lobes were diagnosed with CHP, and the other two patients were diagnosed with IPF.
Many biopsy-proven cases of UIP present on HRCT as a "probable UIP" pattern or even in the presence of inconsistent findings with UIP 16,23 . SLB may not have been necessary in some of these cases, and its avoidance spares patients from associated comorbidities. Some areas of hypoattenuated lung appearing on expiratory imaging, considered to represent air trapping in visual or quantitative analysis, could actually be segments of severely fibrotic non-compliant lung that fail to recoil during expiration. On the other hand, our study findings support consideration of the presence of air trapping in the superior lobes as inconsistent with the UIP pattern and warranting SLB. Moreover, software-assisted quantification of air trapping using the cutoffs employed in this study did not distinguish IPF from other causes of radiological UIP. Howerver, patients with IPF and air trapping, as determined by the E/I ratio, showed a trend toward worse prognosis during follow-up.
Our study has some limitations. First, it likely involved selection bias, as our institution is a local reference center for the diagnosis and management of ILD, especially IPF, which may have increased the prevalence of these diseases relative to other causes of UIP in our sample. For this reason, the sample may not be representative of the overall prevalence of IPF in the general population, although the clinical outcome of our patients resembles those reported previously 24 . Second, the study sample was small (n = 84). Studies with larger samples should be conducted to corroborate our results. Given the limitations inherent to the retrospective nature of this study, some clinical parameters (e.g.; medication use, comorbidities, functional status) for patients at the time of the CT examination could not be retrieved. Third, the quantification of air trapping and distinction of this condition from cystic areas of honeycombing or emphysema can be challenging; for this reason, we did not include areas of attenuation below −950 HU, and we performed visual analysis of the areas of air trapping delimitated by the software when necessary. Lastly, the literature contains no data substantiating the quantitative assessment of air trapping in subjects with IPF. Further studies are need to establish more accurate cutoffs values for air trapping in this population. In conclusion, air trapping is common in patients with ILDs showing the UIP pattern on CT, as determined by qualitative and quantitative evaluation, and it should not be considered to be inconsistent with UIP. On the subjective visual assessment, air trapping in the upper lobes was associated with non-IPF diagnoses.