Identification of malignant ascites using MR-based T1 mapping.

Abstract

e16719 Background: Non-invasive identification of malignant ascites is a challenge in clinical practice.Thus, we decided to assess if an MR-based T1 mapping approach allows non-invasive differentiation of malignant and non-malignant effusions. Methods: In-vitro and ex-vivo MR-examinations were performed on a clinical 1.5T MR-system. T1 mapping was performed with spectroscopy and an adapted modified Look-Locker inversion-recovery (MOLLI) acquisition. For in-vitro experiments 13 titrated solutions with varying albumin content (0 to 200 g/l) were examined. For ex-vivo evaluation 27 ascites/pleural effusion samples from patients with malignancy (19 with histologic tumor confirmation in effusion) and 18 samples from patients without malignancy were examined. All samples underwent histological and laboratory testing. Samples were classified as malignant-positive histology, malignant-negative histology and non-malignant negative histology. Lab values were correlated with T1 maps and receiver operating characteristic (ROC) analysis was used to determine the optimal T1-value threshold to differentiate malignant and non-malignant ascites. Results: In in-vitro analysis both methods showed a high correlation with albumin-content (MOLLI: r = -0.97; Spectroscopy: -0.98). T1-values derived from the reference standard (Spectroscopy) and the MOLLI technique had a high agreement (intraclass correlation single measures: 0,9889, 95% CI: 0,52 to 0,99; average measures: 9,994, 95% CI 0,69 to 0,99). Bland-Altman analysis showed a strong agreement between both methods: 62.5 ± 35 (95% CI: 41.2 to 83.8) Ex-vivo analysis revealed significant differences between T1 values from patients with malignant+ histology (median: 2237; IQR: 2132 to 2327.5) and patients with non malignant- negative histology (median: 2611; IQR: 2548 to 2803, p < 0.0001) as well as between malignant+ histology and all other included patients (median: 2585; IQR: 2503 to 2710, p < 0.0001) Multiple regression analysis of in-vivo results revealed that only albumin content correlated with MOLLI based T1 measurements (p < 0.0001; r = -0.65) ROC analysis for differentiation between malignant and non-malignant effusions (malignant+ histology vs. all other) showed an AUC of 0.897; 95% CI: 0.769 to 0.967). Malignant+ histology vs. non-malignant- histology showed an AUC of 1.000 (cut off Lolli > 2419; 95% CI: 0.905 to 1). Conclusions: T1 Mapping shows excellent correlation with protein content of fluids.MR- T1 mapping allows for non-invasive differentiation of malignant and non-malignant effusions in an ex-vivo set up.