In this study, we retrospectively took into consideration the LGE-MRI scans acquired between April 2016 and March 2019 in 20 patients with CPFE suspected at HRCT. Patients were selected if there was difficulty in differentiating fibrosis from emphysema on HRCT in a multidisciplinary conference setting. This typically arose where lower lobe cysts were subpleural in location, had perceptible walls, and were round and clustered, features that are frequently seen in both fibrosis and emphysema [9]. Exclusion criteria included an acute chest infection or acute exacerbation of lung disease at the time of the investigations, contraindications to gadolinium-based contrast agents such as renal failure, and contraindications to MRI such as severe claustrophobia. The hospital Ethics Board approved the study and written informed consent was waived for this retrospective study. Data from twelve healthy control subjects from a previous study [15] who underwent LGE-MRI were included in the analysis for comparison. Body mass index was recorded for all subjects. Spirometry results were recorded when available. A flow chart diagram is provided in Fig. 1.
High-resolution CT protocol
HRCT scans were acquired on a Siemens Sensation 64-slice CT (Siemens, Erlangen, Germany) from the apices to the costophrenic angles at full inspiration at 120 kVp and 130 mAs. All patients were scanned in the supine position. Contiguous slices were reconstructed at 1-mm slice thickness with a 0.5-mm increment and a 512 × 512 matrix. Automatic tube current modulation was utilised for all patients. Images were reconstructed using lung windows (window width 1,500, centre -700). Images were transferred to a workstation and analysed using quantitative lung analysis software (Pulmo3D, Syngo Via, Siemens, Erlangen, Germany) for calculation of total lung capacity, mean lung density, and the total low attenuation value % (LAV%) of the lungs. For the LAV% an established threshold of -950 HU was used to indicate emphysema [16].
Thoracic MRI protocol
MRI scans were acquired on a 1.5-T magnet (Signa 1.5 T HDX, General Electric Healthcare, Milwaukee, Wis, USA) with an eight-element phased-array cardiac coil (General Electric Healthcare, Aurora, Ohio, USA). After a localising set of axial two-dimensional steady-state-free-precession images were acquired, a bolus injection of 0.2 mmol/kg of gadoterate meglumine (Dotarem, Gd-DOTA, Guerbet, Paris, France) was given, followed by a saline chaser bolus of 20 mL. This was followed 10 min later by a set of axial three-dimensional electrocardiographically gated segmented inversion-recovery prepared fast gradient-echo pulse sequences [17]. The 10-min delay was adapted from previously well-validated cardiac MRI protocols assessing myocardial fibrosis [18, 19]. Sequence parameters were as follows: matrix, 224 × 128; sensitivity encoding factor, 2; inversion time, 130–260 ms (individually optimised to null pulmonary artery blood signal); flip angle, 10°; in-plane resolution, 1.5 × 1.5 mm2; no slice gap; section thickness 1.5 mm resulting in a craniocaudal volume covering 2.5 cm per acquisition. Median number of acquisitions to cover the lungs from apex to base was 7 (interquartile range [IQR] 6–11, depending on the height of the patient). Each acquisition required a 10–15-s breath-hold. In order to null the contrast signal within the pulmonary circulation, inversion times were chosen based on the inversion time to null the blood pool in the pulmonary artery. We assessed the pulmonary artery null-time and adjusted accordingly for each acquisition, adapting similar principles from previously validated cardiac MRI approaches [20]. For patients unable to adequately breath-hold, we used nasal prongs oxygen delivery and a reduction in phase encoding steps to reduce sequence acquisition time. Scan time was recorded from the start to the end of image acquisition. Any complications such as contrast allergy were recorded.
Image interpretation
Two radiologists with 15 and 2 years of experience read all scans in consensus and in random order, blinded to all clinical data. The LGE-MRI images were read blinded to the HRCT images and vice versa. For LGE-MRI, the signal intensity (SI) of fibrosis was measured by placing regions of interest (ROI) in regions of contrast-enhancement in CPFE patients. The ROIs were sized to avoid adjacent non-contrast lung and pulmonary vascularity. A similar process was performed for the HRCT scans. The SI of similar regions of lung in control subjects was also recorded for comparison. The SI of regions of emphysema was measured by placing ROIs in the right and left upper lobes in CPFE patients. A similar process was performed for the HRCT scans. The SI of similar regions of lung in control subjects was also recorded for comparison. LGE was defined using a SI threshold of > 3 standard deviations above the median SI of the upper lobes of the control subjects. The standard deviation of image noise was measured in an ROI outside the body to assess the contrast to noise ratio. The contrast-to-noise ratio was calculated by (median SI of the lung region of elevated SI − median SI of the upper lobe lung region)/1.5 × (standard deviation of noise), as used by others [18]. The percentage SI within pulmonary fibrosis relative to emphysema was calculated as 100 × (median SI of high SI lung region − median SI of upper lobe lung region)/(median SI of upper lobe lung region, adapted from previous myocardial fibrosis techniques [18]. The scoring system by Salisbury et al. [21] was adapted to score features of pulmonary fibrosis. Briefly, readers semiquantitatively scored the presence and extent of (i) reticulation, (ii) honeycombing, and (iii) traction bronchiectasis as follows: 0 (0% involvement of the lung with the feature), 1 (1–10% involvement), 2 (11–20%), 3 (21–30%), 4 (31–40%), 5 (41–50%); 6 (51–60%), 7 (61–70%), 8 (71–80%), 9 (81–90%), and 10 (91–100%). The midpoint of each category was arbitrarily used to convert the semiquantitative (0–10) to quantitative (0–100%) scores. For each scan, the percent involvement of each category for the right and left lungs were averaged to obtain a total lung score, and the radiologists’ total lung scores were averaged to obtain total volume occupied by each feature (HRCT was the reference standard in this study). The Fleischner society nomenclature recommendations were used to define abnormalities [22]. Scan quality for CT and MRI was scored as excellent, good, poor, or uninterpretable.
Statistical analysis
Categorical data were expressed as number and percentage and continuous data as median and IQR. Comparisons of nominal values were performed using the χ2 test, and for scale values, the independent t test was used. Receiver operator characteristic (ROC) analysis was used to analyse the SI difference between regions of pulmonary fibrosis in CPFE patients with similar regions in control subjects to differentiate regions of pulmonary fibrosis from regions of dependent atelectasis (a p value was generated comparing the calculated AUC in the study to a hypothetical chance level AUC [= 0.5]). Sensitivity, specificity, positive and negative predictive values of LGE-MRI in characterising reticulation, honeycombing, traction bronchiectasis, and any fibrosis were calculated using HRCT as the reference standard.
All analyses were carried out using SPSS statistical software (version 13, SPSS Inc., Chicago, IL). A p value < 0.05 was considered to indicate a statistically significant result.