Sebastian Niko Nagel1, Damon Kim2, Tobias Penzkofer3, Ingo G Steffen4, Sebastian Wyschkon5, Bernd Hamm6, Stefan Schwartz7, Thomas Elgeti8. 1. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: sebastian.nagel@charite.de. 2. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany; Institut für Röntgendiagnostik, HELIOS Klinikum Berlin-Buch, Schwanebecker Chaussee 50, 13125 Berlin, Germany. Electronic address: damon.kim@charite.de. 3. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: tobias.penzkofer@charite.de. 4. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: ingo.steffen@charite.de. 5. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: sebastian.wyschkon@charite.de. 6. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: bernd.hamm@charite.de. 7. Medizinische Klinik mit Schwerpunkt Hämatologie und Onkologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: stefan.schwartz@charite.de. 8. Klinik und Hochschulambulanz für Radiologie, Charité - Universitätsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12203 Berlin, Germany. Electronic address: thomas.elgeti@charite.de.
Abstract
OBJECTIVE: To investigate 3T pulmonary magnetic resonance imaging (MRI) for characterization of solid pulmonary lesions in immunocompromised patients and to differentiate infectious from malignant lesions. MATERIALS AND METHODS: Thirty-eight pulmonary lesions in 29 patients were evaluated. Seventeen patients were immunocompromised (11 infections and 6 lymphomas) and 12 served as controls (4 bacterial pneumonias, 8 solid tumors). Ten of the 15 infections were acute. Signal intensities (SI) were measured in the lesion, chest wall muscle, and subcutaneous fat. Scaled SIs as Non-enhanced Imaging Characterization Quotients ((SILesion-SIMuscle)/(SIFat-SIMuscle)*100) were calculated from the T2-weighted images using the mean SI (T2-NICQmean) or the 90th percentile of SI (T2-NICQ90th) of the lesion. Simple quotients were calculated by dividing the SI of the lesion by the SI of chest wall muscle (e.g. T1-Qmean: SILesion/SIMuscle). RESULTS: Infectious pulmonary lesions showed a higher T2-NICQmean (40.1 [14.6-56.0] vs. 20.9 [2.4-30.1], p<0.05) and T2-NICQ90th (74.3 [43.8-91.6] vs. 38.5 [15.8-48.1], p<0.01) than malignant lesions. T1-Qmean was higher in malignant lesions (0.85 [0.68-0.94] vs. 0.93 [0.87-1.09], p<0.05). Considering infections only, T2-NICQ90th was lower when anti-infectious treatment was administered >24h prior to MRI (81.8 [71.8-97.6] vs. 41.4 [26.6-51.1], p<0.01). Using Youden's index (YI), the optimal cutoff to differentiate infectious from malignant lesions was 43.1 for T2-NICQmean (YI=0.42, 0.47 sensitivity, 0.95 specificity) and 55.5 for T2-NICQ90th (YI=0.61, 0.71 sensitivity, 0.91 specificity). Combining T2-NICQ90th and T1-Qmean increased diagnostic performance (YI=0.72, 0.77 sensitivity, 0.95 specificity). CONCLUSION: Considering each quotient alone, T2-NICQ90th showed the best diagnostic performance and could allow differentiation of acute infectious from malignant pulmonary lesions with high specificity. Combining T2-NICQ90th with T1-Qmean increased overall performance, especially regarding sensitivity.
OBJECTIVE: To investigate 3T pulmonary magnetic resonance imaging (MRI) for characterization of solid pulmonary lesions in immunocompromised patients and to differentiate infectious from malignant lesions. MATERIALS AND METHODS: Thirty-eight pulmonary lesions in 29 patients were evaluated. Seventeen patients were immunocompromised (11 infections and 6 lymphomas) and 12 served as controls (4 bacterial pneumonias, 8 solid tumors). Ten of the 15 infections were acute. Signal intensities (SI) were measured in the lesion, chest wall muscle, and subcutaneous fat. Scaled SIs as Non-enhanced Imaging Characterization Quotients ((SILesion-SIMuscle)/(SIFat-SIMuscle)*100) were calculated from the T2-weighted images using the mean SI (T2-NICQmean) or the 90th percentile of SI (T2-NICQ90th) of the lesion. Simple quotients were calculated by dividing the SI of the lesion by the SI of chest wall muscle (e.g. T1-Qmean: SILesion/SIMuscle). RESULTS: Infectious pulmonary lesions showed a higher T2-NICQmean (40.1 [14.6-56.0] vs. 20.9 [2.4-30.1], p<0.05) and T2-NICQ90th (74.3 [43.8-91.6] vs. 38.5 [15.8-48.1], p<0.01) than malignant lesions. T1-Qmean was higher in malignant lesions (0.85 [0.68-0.94] vs. 0.93 [0.87-1.09], p<0.05). Considering infections only, T2-NICQ90th was lower when anti-infectious treatment was administered >24h prior to MRI (81.8 [71.8-97.6] vs. 41.4 [26.6-51.1], p<0.01). Using Youden's index (YI), the optimal cutoff to differentiate infectious from malignant lesions was 43.1 for T2-NICQmean (YI=0.42, 0.47 sensitivity, 0.95 specificity) and 55.5 for T2-NICQ90th (YI=0.61, 0.71 sensitivity, 0.91 specificity). Combining T2-NICQ90th and T1-Qmean increased diagnostic performance (YI=0.72, 0.77 sensitivity, 0.95 specificity). CONCLUSION: Considering each quotient alone, T2-NICQ90th showed the best diagnostic performance and could allow differentiation of acute infectious from malignant pulmonary lesions with high specificity. Combining T2-NICQ90th with T1-Qmean increased overall performance, especially regarding sensitivity.
Authors: Damon Kim; Thomas Elgeti; Tobias Penzkofer; Ingo G Steffen; Laura J Jensen; Stefan Schwartz; Bernd Hamm; Sebastian N Nagel Journal: Eur Radiol Date: 2020-08-21 Impact factor: 5.315