MRI Compared with Low-Dose CT for Incidental Lung Nodule Detection in COPD: A Multicenter Trial

Abstract

Purpose

To investigate morphofunctional chest MRI for the detection and management of incidental pulmonary nodules in participants with chronic obstructive pulmonary disease (COPD).

Materials and Methods

In this prospective study, 567 participants (mean age, 66 years ± 9 [SD]; 340 men) underwent same-day contrast-enhanced MRI and nonenhanced low-dose CT (LDCT) in a nationwide multicenter trial (clinicaltrials.gov: NCT01245933). Nodule dimensions, morphologic features, and Lung Imaging Reporting and Data System (Lung-RADS) category were assessed at MRI by two blinded radiologists, and consensual LDCT results served as the reference standard. Comparisons were performed using the Student t test, and agreements were assessed using the Cohen weighted κ.

Results

A total of 525 nodules larger than 3 mm in diameter were detected at LDCT in 178 participants, with a mean diameter of 7.2 mm ± 6.1 (range, 3.1–63.1 mm). Nodules were not detected in the remaining 389 participants. Sensitivity and positive predictive values with MRI for readers 1 and 2, respectively, were 63.0% and 84.8% and 60.2% and 83.9% for solid nodules (n = 495), 17.6% and 75.0% and 17.6% and 60.0% for part-solid nodules (n = 17), and 7.7% and 100% and 7.7% and 50.0% for ground-glass nodules (n = 13). For nodules 6 mm or greater in diameter, sensitivity and positive predictive values were 73.3% and 92.2% for reader 1 and 71.4% and 93.2% for reader 2, respectively. Readers underestimated the long-axis diameter at MRI by 0.5 mm ± 1.7 (reader 1) and 0.5 mm ± 1.5 (reader 2) compared with LDCT (P < .001). For Lung-RADS categorization per nodule using MRI, there was substantial to perfect interreader agreement (κ = 0.75–1.00) and intermethod agreement compared with LDCT (κ = 0.70–1.00 and 0.69–1.00).

Conclusion

In a multicenter setting, morphofunctional MRI showed moderate sensitivity for detection of incidental pulmonary nodules in participants with COPD but high agreement with LDCT for Lung-RADS classification of nodules.

Clinical trial registration no. NCT01245933 and NCT02629432

Keywords: MRI, CT, Thorax, Lung, Chronic Obstructive Pulmonary Disease, Screening

© RSNA, 2023

Supplemental material is available for this article.

Summary

In this multicenter trial, morphofunctional MRI demonstrated potential as an alternative screening modality to low-dose CT to aid in detection and Lung Imaging Reporting and Data System classification of incidental pulmonary nodules in participants with chronic obstructive pulmonary disease.

Key Points

• ■ In this German multicenter trial of 567 participants with chronic obstructive pulmonary disease from 16 centers, morphofunctional MRI achieved 71.4%–73.3% sensitivity for detection of incidental pulmonary nodules 6 mm or greater in diameter, with same-day low-dose CT (LDCT) as the reference standard.

• ■ For Lung Imaging Reporting and Data System (Lung-RADS) classification per nodule using MRI, there was substantial intermethod agreement (MRI vs LDCT) (κ = 0.68 and 0.65, respectively) and interreader agreement (κ = 0.76).

• ■ For Lung-RADS classification per participant using MRI, there was substantial intermethod agreement (κ = 0.79 and 0.76, respectively) and almost-perfect interreader agreement (κ = 0.88).

Introduction

Morphofunctional MRI has recently been introduced as an alternative modality to chest CT for imaging muco-obstructive lung diseases, such as cystic fibrosis and chronic obstructive pulmonary disease (COPD) (1–4). In addition to the benefit of no ionizing radiation, MRI provides several functional techniques for imaging these diseases. Pulmonary nodules and lung cancer are important comorbidities of COPD, and routine cross-sectional imaging with MRI may depict clinically relevant nodules. Studies in high-risk populations indicated a 33% (range, 17%–53%) incidence of pulmonary nodules, and widespread application of low-dose CT (LDCT) for screening demonstrates that COPD may be the single most powerful predictor of lung cancer, whether smoking-related or not (5,6).

The German nationwide multicenter trial COSYCONET (COPD and SYstemic consequences-COmorbidities NETwork) performs comprehensive clinical and functional assessments of patients at risk for and diagnosed with COPD and provides the unique opportunity to compare morphofunctional MRI with same-day LDCT for the detection and characterization of incidental pulmonary nodules in a nonscreening setting of COPD (7). The Lung Imaging Reporting and Data System (Lung-RADS) was developed to guide management of screening-detected nodules in a highly standardized fashion, but its applicability to MRI-detected nodules is largely unstudied. Furthermore, nodule detection and categorization with morphofunctional MRI has not been assessed in a multicenter approach (8–10). Lung-RADS is based mainly on nodule dimension measurements and few morphologic criteria. At present, a threshold of 6 mm in diameter is considered sufficient to initiate further follow-up, and solid nodules 15 mm or greater or a solid component 8 mm or greater within part-solid nodules indicate the need for immediate workup (11).

The present study aimed to assess the sensitivity of multicenter morphofunctional chest MRI at 1.5 T and 3.0 T for the detection of incidental pulmonary nodules 3 mm or greater in diameter in patients with COPD from the COSYCONET study and investigate the ability of MRI to guide nodule management based on Lung-RADS recommendations, using a noninferiority approach, in comparison with nonenhanced LDCT.

Materials and Methods

Study Design

The study is part of the prospective longitudinal multicenter COPD cohort trial titled Impact of Systemic Manifestations/Comorbidities on Clinical State, Prognosis, Utilisation of Health Care Resources in Patients with COPD (COSYCONET; Clinicaltrials.gov: NCT01245933) and the imaging substudy Image-Based Structural and Functional Phenotyping of the COSYCONET Cohort Using MRI and CT (MR-COPD; Clinicaltrials.gov: NCT02629432). Each center obtained approval from the local ethics committee as well the German Federal Office for Radiation Protection, and all participants gave their written informed consent.

The precise inclusion and exclusion criteria of MR-COPD can be found elsewhere (7,12–14). In short, individuals 40 years or older with COPD diagnosed according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria, as well as individuals with no assignable GOLD category (including the former GOLD 0 category), were evaluated for inclusion as individuals at risk (12,15). Exclusion criteria were previous major lung surgery, moderate to severe exacerbations within the last 4 weeks, active pneumonia, and contraindications to CT or MRI.

A total of 2741 participants were recruited in 31 study centers throughout Germany for the COSYCONET trial, of whom 607 were prospectively recruited for the MR-COPD imaging substudy, comprising 16 centers. Thirty-five participants were excluded because of incomplete examinations, and five were removed from the analysis because of severe respiratory artifacts that impaired image analysis, resulting in a total of 567 participants in the final analysis (Fig 1).

Figure 1:

Study flowchart. COPD = chronic obstructive pulmonary disease, Gd = gadolinium-based, GOLD = Global Initiative for Chronic Obstructive Lung Disease, LDCT = low-dose CT.

Of note, the COSYCONET consortium has reported numerous articles based on this large multicenter trial of COPD comorbidities, treatments, spirometry results, and quantitative MRI and LDCT findings. The manuscripts associated with COSYCONET are listed elsewhere (https://h009.ssl-redirect.de/www.asconet.net/edit/cosyconet/publik/manus). Specifically, functional quantitative MRI and LDCT data have been compared in a recent study using only a small subgroup of patients (12). Here, we report data from the only study on incidental pulmonary nodules in the COSYCONET imaging substudy.

MRI Protocol

Morphofunctional MRI was performed with 1.5-T and 3.0-T clinical scanners as previously described, comprising steady-state free precession, gradient-echo, and half-Fourier single-shot turbo spin-echo acquisitions (eg, HASTE by Siemens) acquired in inspiratory and expiratory breath hold in transversal and coronal orientations (Tables S1, S2) (12–14). For perfusion MRI, a bolus of gadobutrol (Gadavist; Bayer Healthcare), calculated as 0.1 mmol per kilogram of body weight, was injected at a rate of 3–5 mL/sec using a power injector, followed by a saline flush of 20 mL. The gradient-echo acquisitions were repeated after administration of contrast material in both inspiration and expiration, with additional fat saturation preparation for the transversal images. Breath-holding times were less than 20 seconds. The total acquisition time was approximately 30 minutes. The MRI protocol was controlled regularly by a custom-designed phantom on every machine (13).

LDCT Protocol

LDCT was performed with at least 40 detector-row scanners from different manufacturers (Siemens Healthineers, General Electric, Philips). The LDCT protocol was defined according to the recommendations of the German Radiological Society and was composed of helical acquisitions in inspiration and end expiration (Table S3). CT images of 0.625–1.0-mm contiguous section thickness were reconstructed with smooth and edge-enhancing algorithms (B70f/LUNG/L and B30f/SOFT/B, generic names of Siemens/General Electric/Philips). The maximum effective dose of paired LDCT was less than 3.5 mSv, which is similar to published doses in the context of lung cancer screening (16). The LDCT protocol was validated by repeated phantom measurements on every scanner (17).

Image Assessment

MRI and LDCT images were reviewed independently by two general radiologists blinded to each other’s results and all clinical data (Q.L. and L.Z., with 7 and 3 years of experience, respectively) using a standardized questionnaire comprising detection, axis measurements, and morphologic details, consistent with previous studies (Fig S1) (18–20). MRI scans were reviewed first and at least 1 month before LDCT to prevent memory bias. Image assessment was performed using a dedicated workstation based on an iMac 27-inch screen (Apple) with two radiology-grade viewing screens running a picture archiving and communication system (Horos, version 3.3.1; Horos Project). According to previous reports and in keeping with clinical guidelines for nodules requiring medical attention, nodules smaller than 3 mm were not recorded (18,21–23). The same assessments were performed for both MRI and LDCT, including longest axial diameter and perpendicular short-axis diameter of each nodule. Attenuation was rated in three categories as solid, part-solid, or purely ground-glass opacity (GGO). In case of a part-solid nodule, the percentage of GGO component of the whole nodule was rated in four categories: 0%–25%, 26%–50%, 51%–75%, and 76%–100% (18). Morphologic characteristics, such as calcification, fat, spiculated or lobulated margins, bronchial cutoff sign, and cavitation, were recorded (18,20). Finally, the Lung-RADS score was assigned to each nodule (11). The axis measurements for MRI were performed on transversal inspiratory contrast-enhanced T1-weighted three-dimensional gradient-echo images and for LDCT on inspiratory transversal reconstructions with lung kernel and window. For LDCT, values of axis measurements from both readers were averaged, and the variables of detection and categorization were re-evaluated in consensus by the two readers when initial rating was different to establish a consensus for LDCT as a reference standard.

Statistical Analysis

Statistical analyses were performed with SPSS software, version 18 (IBM), and SigmaPlot, version 12.5 (Systat Software), by an author (Q.L.). Data are presented as means ± SDs, and comparisons were performed with a Student t test unless otherwise specified. LDCT consensual results served as the standard of reference for nodule attenuation type and size, calculation of parameters of diagnostic performance (sensitivity, specificity, positive predictive value), and Lung-RADS categorization of MRI. In a per-nodule approach, nodules were considered true-positive if identified at MRI and confirmed with LDCT in consensus of both readers and false-positive when identified at MRI but not detected with LDCT. A false-negative result is defined by a nodule not identified at MRI but present at LDCT. In a per-participant approach, a true-positive result indicates participants with at least one nodule identified with MRI and confirmed with LDCT, whereas a false-positive result indicates participants with nodules apparent at MRI but not at LDCT. A true-negative result indicates participants without any nodule at MRI and LDCT, and false-negative indicates participants with nodule detected with LDCT but not detected with MRI. In the per-participant approach, only the most suspicious nodule was selected in participants with multiple nodules to determine further management based on Lung-RADS, and those without nodules (including false-negative results for MRI) were rated as Lung-RADS category 1. To compare differences in long- and short-axis measurements between MRI and LDCT, Bland-Altman analysis was applied, and the measurement associations were assessed by Pearson correlation coefficient. Interreader and intermethod agreements were assessed with Cohen weighted κ values (24), which were classified as poor (κ = 0–0.20), fair (κ = 0.21–0.40), moderate (κ = 0.41–0.60), substantial (κ = 0.61–0.80), and almost perfect (κ = 0.81–1.00) agreement, as previously described (25). A P value less than .05 indicated a statistically significant difference.

Results

Participant Characteristics

A total of 567 participants (mean age, 66 years ± 9 [SD] [range, 42–85 years]; 340 men and 227 women) were included (Fig 1). Of these, 81 (14.3%) participants were categorized as GOLD stage I, 234 (41.3%) as GOLD II, 129 (22.7%) as GOLD III, and 27 (4.8%) as GOLD IV; 96 (16.9%) did not have airflow impairment but were at risk for COPD (former Gold 0) (Table 1).

Table 1:

Participant Characteristics

Sensitivity of Morphofunctional MRI for Incidental Pulmonary Nodules

LDCT depicted 525 nodules in 178 of 567 (31.4%) participants: Of these, 495 (94.3%) nodules were solid, 17 (3.2%) were part-solid, and 13 (2.5%) were purely GGO. Overall, readers 1 and 2 identified 316 of 525 (60.2%) and 302 of 525 (57.5%) nodules at MRI, respectively (Table 2; Figs 2, 3), with substantial overall interreader agreement (κ = 0.80; 95% CI: 0.75, 0.85). Regarding nodule size, MRI yielded low sensitivity for solid nodules less than 6 mm in diameter: 52.6% (161 of 306) and 49.0% (150 of 306) for readers 1 and 2, respectively; sensitivity was higher for nodules 6 mm or greater: 73.3% (154 of 210) for reader 1 and 71.4% (150 of 210) for reader 2 (Table 2). MRI demonstrated positive predictive values of 78.5% (reader 1) and 75.8% (reader 2) for nodules less than 6 mm and of 92.2% and 93.2%, respectively, for nodules 6 mm or greater (Table 2). Sensitivity for part-solid and GGO nodules was low irrespective of nodule size, which may be due in part to low prevalence; only six of 178 (3.4%) and two of 178 (1.1%) participants had part-solid and GGO nodules, respectively.

Table 2:

Diagnostic Performance of Morphofunctional MRI for Pulmonary Nodules Grouped by Long-Axis Diameter

Figure 2:

Representative images in axial plane of a solid nodule at morphofunctional MRI and low-dose CT (LDCT). (A) A 70-year-old woman with a solid pulmonary nodule (white arrow) in left lower lobe at LDCT. (B–D) The same solid nodule (white arrow) shown on (B) T2-weighted MR images and on (C) nonenhanced and (D) contrast-enhanced T1-weighted images. Spiculation can be identified at LDCT and MRI.

Figure 3:

Representative images in axial plane of a part-solid nodule at morphofunctional MRI and low-dose CT (LDCT). (A) A 72-year-old woman with a part-solid nodule in right upper lobe at LDCT. (B–D) Nodule was correctly identified on (B) T2-weighted and (C) nonenhanced and (D) contrast-enhanced T1-weighted MR images. The maximum axial diameter of the solid component (black arrowhead in A and D) measured 14.3 mm with MRI and 14.7 mm with LDCT, whereas the extent of the ground-glass opacity component (white arrowheads in A and D) was underestimated at MRI, with a measurement of 31.7 mm, compared with 34.3 mm at LDCT.

MRI Systematically Underestimates Nodule Diameters

The mean long-axis diameter of the 525 nodules detected with LDCT was 7.2 mm ± 6.1 by consensus (range, 3.1–63.1 mm). In general, long- and short-axis measurements were highly correlated between LDCT and MRI for both readers (r = 0.97–0.98; P < .001). By Bland-Altman analysis, MRI led to underestimates of the long-axis diameter compared with LDCT by 0.5 mm ± 1.7 and 0.5 mm ± 1.5 (P < .001) for readers 1 and 2, respectively. Short-axis diameter was also underestimated at MRI by reader 2 (0.4 mm ± 0.9; P < .001) but not by reader 1 (P = .94) (Fig 4). Interreader agreement analysis showed that both readers disagreed to a low degree but systematically (P < .03 to .001) (Fig 4). Although intermethod differences for long-axis diameter measurements in solid nodules were similar between readers, as described above, there was disagreement in intermethod differences for part-solid nodules (−0.2 mm ± 0.8 [P = .73] for reader 1 and −1.1 mm ± 1.3 [P = .29] for reader 2). GGO components were missed by readers at MRI in two of three part-solid nodules and were therefore miscategorized as solid, and long-axis diameter was underestimated by 2.6 mm (reader 1) and 2.0 mm (reader 2) in evaluations of the correctly identified nodule (Fig 3).

Figure 4:

Long- and short-axis diameter measurements of detected nodules for MRI in comparison with low-dose CT (LDCT). (A–F) Graphs show intermethod comparison of long- and short-axis measurement differences for (A, B) reader 1 and (C, D) reader 2, with MRI versus LDCT as standard of reference, and (E, F) interreader agreement for MRI for reader 1 versus reader 2. Dashed lines denote the limits of agreement (LoA), and solid lines denote the mean. Mean bias and LoA are given at the top of each image, including P values. Note that nodules not detected with MRI are not included in the analysis.

Lung-RADS Classification with MRI

At MRI, morphologic features, such as spiculation, bronchial cutoff sign, and cavitation and/or necrosis, were detected with low sensitivity by both readers (41.7%–77.8%) (Table S4). Of note, all calcifications were missed at MRI. The subsequent Lung-RADS categorization of the nodules detected with MRI revealed substantial intermethod agreement with LDCT for both readers (κ = 0.68 and 0.65, respectively), as well as substantial interreader agreement (κ = 0.76; 95% CI: 0.70, 0.82) (Tables 3, 4). Data on subsolid nodules were too scarce to allow for conclusions.

Table 3:

Contingency Table of Lung-RADS Categories for Nodules Detected at MRI by Reader 1 in Comparison with Low-Dose CT

Table 4:

Contingency Table of Lung-RADS Categories for Nodules Detected at MRI by Reader 2 in Comparison with Low-Dose CT

Of 178 participants with at least one nodule identified with LDCT, 145 (81.5%) were correctly detected at MRI by reader 1 and 144 (80.9%) by reader 2, with high specificity (97.4% [379 of 389] and 96.4% [375 of 389] for readers 1 and 2, respectively) and high interreader agreement (κ = 0.95; 95% CI: 0.92, 0.98). Subsequent per-participant Lung-RADS categorization revealed a somewhat improved intermethod agreement compared with the per-nodule approach, with κ values of 0.79 and 0.76 for readers 1 and 2, respectively, and an almost perfect interreader agreement regarding the most suspicious nodule (κ = 0.88; 95% CI: 0.84, 0.91) (Tables S5, S6).

Discussion

In our multicenter study using LDCT as the reference standard, morphofunctional MRI had 73.3% and 71.4% sensitivity and 92.2% and 93.2% positive predictive value for detecting incidental nodules 6 mm or greater in participants with or at risk for COPD by readers 1 and 2, respectively. Nodule size measurements at MRI were highly correlated with LDCT, but readers systematically underestimated both long- and short-axis diameter measurements at MRI within a range of less than 1 mm. Lung-RADS categorization with MRI was found to be in substantial concordance with LDCT in a per-participant approach. Almost all participants (99%) were able to adhere to the scanning regimen and breath-holding times.

Initial studies on MRI for the assessment of lung nodules in a phantom revealed that the threshold for detecting nodules is likely 3–4 mm (22,26,27). A recent screening trial with 233 patients in a single center showed 69.7% sensitivity for solid nodules smaller than 6 mm and 97.6% sensitivity for nodules 6 mm or greater compared with LDCT using noncontrast T1- and T2-weighted sequences and diffusion-weighted imaging (10). The 6-mm threshold for clinically meaningful solid and/or part-solid nodules in screening trials resulted in further short-term imaging controls or further workup (28). The threshold size in our study was arbitrarily set at 3 mm to test maximal sensitivity of MRI, but additional subgroup analyses for nodules 6 mm or greater were performed. In these participants, MRI achieved a sensitivity of 75.5%–88.5% for solid nodules but low sensitivity for subsolid nodules (13.3%) by both readers. By comparison, Meier-Schroers et al (29) reported a 73% sensitivity of subsolid nodules with MRI. A possible reason for the lower sensitivity achieved for subsolid nodules in our study is sampling bias, as most subsolid nodules were found in the same individual. The present work supports previous studies demonstrating that the ability of MRI to depict small pulmonary nodules less than 6 mm is limited with use of conventional techniques, such as gradient-echo sequences. This limitation is mainly due to the inherent limitations of pulmonary MRI, namely low proton density and frequent tissue–air interfaces; additional conditions, such as hyperinflation, oligemia, and emphysema, in patients with COPD further limit MRI performance (30). Of note, the relatively low detection rate for nodules smaller than 6 mm in our study (which is linked to a low risk of malignancy according to the literature) and low number of false-positive results may have a positive effect on cost-effectiveness by avoiding unnecessary follow-up and interventions (31). Detection of 4–6-mm solid nodules can be expected to improve by using ultrashort echo time MRI, with reported sensitivities of 72%–81% (21).

Several factors influence the ability of MRI to categorize disease. In the present study, readers systemically underestimated the long- and short-axis diameter at MRI by less than 1 mm while maintaining strong correlations between MRI and LDCT, which is in line with previous reports (18). This may have been due to the tendency of MRI to smooth the margins of nodules. Furthermore, as can be observed in Figure 4, the variation of differences increases with nodule dimensions, which may result in an underestimation of the limits of agreement in this study because most nodules were small. This systematic measurement bias influenced Lung-RADS categorization, mostly toward a misclassification in Lung-RADS category 2 (MRI) instead of 3 (LDCT). Of note, these categories differentiate between the need for 6- and 12-month imaging follow-up (11). Because this is a systematic effect and within the range of the usual interreader agreement well known for CT, it may be accounted for when MRI measurements are used for decisions regarding further patient management (32–34). Axis measurements were carried out on T1-weighted gradient-echo images, which provide the highest in-plane resolution of the MRI protocol used. Measuring on images acquired with other sequences may result in larger measurement error compared with LDCT and higher interreader variability. Of note, MRI also tended to underestimate the GGO component in part-solid nodules, which was associated with larger variations compared with the solid component. This interpretation is, however, limited by the low numbers in the present study.

In addition to the low detection ability of MRI for nodules smaller than 6 mm, its insensitivity to calcification also contributes to the tendency toward miscategorization of benign calcified nodules at MRI. On the other hand, Lung-RADS category 4X, which warrants intervention in solid nodules, was underestimated with MRI; this may be attributed to missed morphologic features, such as spiculation. Previous studies suggest that identification of morphologic features can be improved by using ultrashort echo time MRI, with reported sensitivities of 48.0%–61.5% for spiculation and 61.3%–72.7% for internal lucencies, for example (18,35). Subsequently, Ohno et al (35,36) reported an almost perfect agreement (κ = 0.82; P < .001) between ultrashort echo time and LDCT for Lung-RADS classification in a lung cancer screening study.

Remarkably, MRI has come close to the categorization ability of LDCT, achieving substantial intermethod and interreader agreement for Lung-RADS classification of identified nodules. In the per-participant approach considering the most suspicious nodule only, MRI had higher sensitivity and specificity for Lung-RADS categorization compared with the per-nodule approach. Most nodules in our study were low- to intermediate-risk categories (70.9% and 8.2% for Lung-RADS 2 and 3, respectively) because our study sample was not based on a preselection of participants with known nodules. These numbers are similar to the results of a recent MRI screening study of lung nodules by Ohno et al (36), in which 65.8% of nodules were found to be classified as Lung-RADS 2, 20.4% as Lung-RADS 3, and only 13.8% as Lung-RADS 4. Furthermore, 11 of 567 participants in our study were classified as Lung-RADS 4X based on the per-participant approach, which is also similar to the odds of malignancy (approximately 1.9%) found in eight of the largest LDCT-based lung cancer screening trials worldwide (37).

Our study had some limitations. First, LDCT served as the standard of reference, and histologic confirmation is missing. Second, the selection criteria for participation in COSYCONET (age ≥ 40 years, diagnosis of COPD irrespective of smoking status) were different from the selection recommendations for lung cancer screening according to the U.S. Preventive Services Task Force (age 50–80 years with ≥20 pack-year smoking history) (12,15,38). Only 389 of 567 (68.6%) participants did not show nodules greater than 3 mm in our study, which is a lower rate of normal scans compared with National Lung Cancer Screening Trial data (75.8% negative) (39). Because the NLST provided the basis for the Lung-RADS classification system, actual malignancy risk of nodules in our study can be different from Lung-RADS estimates. Future studies will need to investigate the ability of MRI to specifically detect malignant nodules and provide follow-up measurements of nodules. Further, we could not include ultrashort echo time sequences in this multicenter setting because of its low availability and lack of standardization at the time of the study (35).

In conclusion, our study showed promising results of morphofunctional MRI as an alternative screening tool for the detection and classification of incidental pulmonary nodules 6 mm or greater in participants with COPD in a multicenter setting. Our data suggest that a substantial number of pulmonary nodules, a frequent comorbidity of COPD, may be correctly identified with MRI. At present, MRI cannot replace LDCT for lung cancer screening. It is also current practice that nodules observed at MRI are immediately confirmed by using LDCT for patient management decisions. Further follow-up studies should be conducted to assess whether diagnosis and management decisions based on MRI alone are sufficient.