Longitudinal MRI in comparison to low-dose CT for follow-up of incidental pulmonary nodules in patients with COPD-a nationwide multicenter trial

Abstract

Purpose: This multicenter trial was conducted to evaluate MRI for the longitudinal management of incidental pulmonary nodules in heavy smokers.

Materials and methods: 239 participants (63.9 ± 8.4 years, 43-82 years) at risk of or with COPD GOLDI-IV from 16 centers prospectively underwent two rounds of same-day low-dose computed tomography (LDCT1&2) and MRI1&2 at an interval of three years in the nationwide COSYCONET trial. All exams were independently assessed for incidental pulmonary nodules in a standardized fashion by two blinded readers, incl. axis measurements and Lung-RADS categorization, with consensual LDCT results serving as the standard of reference. A change in diameter ≥ 2 mm was rated as progress. 11 patients underwent surgery for suspicious nodules after the first round.

Results: Two hundred twenty-four of two hundred forty nodules (93.3%) persisted from LDCT1 to LDCT2, with a sensitivity of MRI2 of 82.8% and 81.5% for readers 1 and 2, respectively. Agreement in Lung-RADS categories between LDCT2 and MRI2 was substantial in per-nodule (κ = 0.62-0.70) and excellent in a per-patient (κ = 0.86-0.88) approach for both readers, respectively. Concordance between LDCT2 and MRI2 for growth was excellent to almost perfect (κ = 0.88-1.0). The accuracy of LDCT1 and MRI1 for lung cancer was 87.5%. Lung-RADS ≥ 3 category on MRI1 had higher accuracy for predicting progress (23.1% and 21.4%, respectively) than LDCT1 (15.8%).

Conclusion: Compared to LDCT, MRI shows similar capabilities for the longitudinal evaluation of incidental nodules in heavy smokers. Decision-making for nodule management guided by Lung-RADS seems feasible based on longitudinal MRI.

Key points: Question Can MRI serve as an alternative to low-dose CT (LDCT) for the longitudinal management of pulmonary nodules in heavy smokers, addressing concerns over radiation exposure? Findings MRI demonstrated substantial agreement with LDCT in detecting nodule growth, accurately categorizing Lung-RADS, and comparable accuracy in identifying malignancy over a three-year follow-up. Clinical relevance Longitudinal MRI demonstrates high consistency with LDCT in assessing the growth of incidental pulmonary nodules and categorizing per-patient Lung-RADS, offering a reliable, radiation-free alternative for monitoring and early malignancy detection in high-risk populations.

Keywords: Chronic obstructive pulmonary disease; Longitudinal management; Low-dose computed tomography; Magnetic resonance imaging; Pulmonary nodules.

Fig. 1

Study flow chart

Fig. 2

Representative images of progressed and regressed incidental nodules at morpho-functional MRI and LDCT. a A solid pulmonary nodule beside the visceral pleura showed significant growth on the second examination (second row, 2018) compared to the first examination (first row, 2015). The growth can be best displayed on the contrast-enhanced T1-weighted MR images (third column) than the nonenhanced (second column) and T2-weighted images (fourth column). b A solid nodule adjacent to the interlobar pleura was significantly shrunk on the second examination (fourth row) compared to the first round examination (third row). The shrunk nodules on the second imaging round were not displayed on the three MRI sequences

Fig. 3

Longitudinal long- and short-axis diameter measurements of incidental nodules on LDCT for two imaging rounds. Comparison of the long- (a, c) and short-axis (b, d) diameter of incidental nodules for LDCT1 vs LDCT2 for Reader 1 (a, b) and Reader 2 (c, d), respectively. Inter-reader agreement for the longitudinal difference in long- (e) and short-axis (f) diameter measurements from LDCT1 to LDCT2. Dashed lines denote 95% confidence interval, and solid lines the mean. Mean bias and 95% CI are given at the bottom of each image, including p-values

Fig. 4

Longitudinal long- and short-axis diameter measurements of incidental nodules on MRI for two imaging rounds. Comparison of the long- (a, c) and short-axis (b, d) diameter of incidental nodules for MRI1 vs MRI2 for Reader 1 (a, b) and Reader 2 (c, d), respectively. Inter-reader agreement for the longitudinal change in long- (e) and short-axis (f) diameter measurements from MRI1 to MRI2. Inter-methods agreement for the average longitudinal change in long- (g) and short-axis (h) diameter measurements by two readers from MRI1 to MRI2 vs LDCT1 to LDCT2 as the standard of reference. Dashed lines denote 95% confidence interval, and solid lines the mean. Mean bias and 95% CI are given at the bottom of each image, including p-values

Fig. 5

Examples of Lung-RADS classification of incidental nodules representing histologically proven lung cancer at LDCT and morpho-functional MRI. a A solid nodule with spiculated edges was classified as Lung-RADS grade 4A on both LDCT1 and MRI1. b A solid mass with lobulated edges and necrosis was classified as Lung-RADS grade 4X on both LDCT1 and MRI1