247Characterisation of pleural and pericardial effusions with T1 mapping

Abstract

Characterisation of pleural and pericardial effusions with T1 mapping Rosmini S.1 ; Seraphim A.1 ; Captur G.1 ; Gomes AC.1 ; Zemrak F.1 ; Treibel TA.1 ; Cash L.1 ; Culotta V.1 ; O"mahony C.1 ; Kellman P.2 ; Moon JC.1 ; Manisty C.1 1Barts Health NHS Trust, Cardiac Imaging Department, London, United Kingdom of Great Britain & Northern Ireland 2National Institutes of Health, Bethesda, United States of America Background: Determining the cause of pleural and pericardial effusions currently involves biochemical analysis of fluid obtained invasively and classification as exudates or transudates using the Light’s criteria, known to have reduced specificity. Purpose: We hypothesised that T1 mapping by Cardiovascular Magnetic Resonance (CMR) may aid differentiation between exudative and transudative effusions. Methods: A phantom was created containing variable albumin concentrations (25g/L, 50g/L, 100g/L, 200g/L)(Figure 1A) to study the relationship between fluid T1 and protein concentration. Imaging at 1.5T (Siemens Aera) with T1 mapping using single-shot steady-state freeprecession (SSFP)-based MOdified Look-Locker Inversion Recovery (MOLLI, 5s(3s)3s) showed close inverse correlation between protein concentration and T1 values (r=-0.992),Figure 1B. 64 patients were then evaluated with pleural and/or pericardial effusions who underwent CMR at 1.5T with mapping. Biochemical fluid analysis from thoracocentesis and/or pericardiocentesis was available for 37 patients. The effusions of the remaining 27 were classified as transudates based on unequivocal diagnoses of chronic severe left ventricular (LV) systolic dysfunction (SD)(ejection fraction < 30%). T1 maps were acquired, and mean fluid T1 measured to evaluate its diagnostic performance for differentiation between exudates and transudates. Results: T1 values were significantly lower with exudate (n = 9) than transudate (n = 29) pleural effusions (3317 ± 241ms vs 3581 ± 222ms, p = 0.004), and with pericardial effusions (exudates, n= 19, 2809± 399ms vs transudates, n= 17, 3337 ± 246ms, p < 0.0001)(Figure 1C,1D,1F). A cut-off T1 value of 3440ms diagnosed transudative pleural effusions with 79% sensitivity, 78% specificity, AUC 0.79, (95% CI 0.63-0.96, p = 0.008). For pericardial effusions, T1 cut-off for transudates was lower - 3013ms (sensitivity 94%, specificity 79%, AUC for 0.86, (95% CI 0.73-0.99), p < 0.0001). For patients with biochemical fluid analysis, T1 values and fluid:plasma protein ratio negatively correlated (pleural: r=-0.630, p = 0.005; pericardial: r=-0.449, p = 0.036). Also T1 of pleural and pericardial exudates (p = 0.002) and of pleural and pericardial transudates (p = 0.001) was significantly different. One patient had haemopericardium following complicated myocardial biopsy(1E). Conclusion: T1 mapping provides non-invasive information about the molecular composition of pericardial and pleural effusions, with diagnostic performance for differentiating transudates from exudates approaching that of Light’s criteria, and can identify haemo-effusions. Figure 1: 1A:protein phantom T1 MOLLI image with good correlation between T1 values and albumin concentration(1B); 1C:pericardial exudate in patient with active fibrinous pericarditis; 1D:pericardial and pleural transudates in patient with severe LVSD; 1E:haemopericardium; 1F:pericardial exudate and pleural transudate in patient with acute lymphoblastic lymphoma.