Evaluation of malignant effusions using MR-based T1 mapping

Our aim was to investigate the diagnostic yield of rapid T1-mapping for the differentiation of malignant and non-malignant effusions in an ex-vivo set up. T1-mapping was performed with a fast modified Look-Locker inversion-recovery (MOLLI) acquisition and a combined turbo spin-echo and inversion-recovery sequence (TMIX) as reference. A total of 13 titrated albumin-solutions as well as 48 samples (29 ascites/pleural effusions from patients with malignancy; 19 from patients without malignancy) were examined. Samples were classified as malignant-positive histology, malignant-negative histology and non-malignant negative histology. In phantom analysis both mapping techniques correlated with albumin-content (MOLLI: r = − 0.97, TMIX: r = − 0.98). MOLLI T1 relaxation times were shorter in malignancy-positive histology fluids (2237 ± 137 ms) than in malignancy-negative histology fluids (2423 ± 357 ms) as well as than in non-malignant-negative histology fluids (2651 ± 139 ms); post hoc test for all intergroup comparisons: < 0.05. ROC analysis for differentiation between malignant and non-malignant effusions (malignant positive histology vs. all other) showed an (AUC) of 0.89 (95% CI 0.77–0.96). T1 mapping allows for non-invasive differentiation of malignant and non-malignant effusions in an ex-vivo set up.

www.nature.com/scientificreports/ dedicated microscopic and cytology analysis was performed on the drawn samples. Samples were classified as malignancy-positive histology if samples were drawn from patients with malignancy and histologic/cytologic analysis revealed malignant cells; as malignancy-negative histology if samples were drawn from patients with malignancy but histologic/cytologic analysis did not reveal malignant cells; and as non-malignant negative histology if samples were drawn from patients without any known malignancies and histologic/cytologic analysis did not reveal malignant cells. The local institutional review board (Ethics Committee of the Faculty of Medicine of Bonn University) approved this study; ex-vivo analysis of effusion samples was exempt from IRB review, a waiver for informed patient consent was approved. All experiments were performed in accordance with relevant guidelines and regulations.
Examinations were performed on albumin emulsions (phantom; predefined albumin content) and on drawn samples of pleural and ascitic fluids (ex-vivo) to determine the impact of albumin content on T1-relaxation time as well as to evaluate whether a differentiation of malignant/non-malignant effusions is possible by means of T1-mapping. MR-testing of samples was performed after these had adapted to room temperature (21 °C).
Phantom analysis. MRI evaluation was performed on a total of 13 albumin emulsions (Human Albumin, Behring, Germany) with varying concentrations of albumin (between 0 and 200 g/l) diluted in sodium chloride (combined fluid volume in all samples was 50 ml).
Statistical analysis. Statistical analyses were performed using commercially available statistical software (Prism version 8, GraphPad, La Jolla, CA). Bland Altman analysis was performed for phantom analyses to determine the agreement between MOLLI and TMIX measurements of T1 relaxation times. Correlation between laboratory protein concentrations and T1 relaxation times was analyzed with Pearson's correlation. Regression analysis was performed to determine which factors (triglyceride levels, protein content, RBC, WBC) were predictive of T1 relaxation times for ex-vivo analysis. Receiver operating characteristic (ROC) curves were plotted for ex-vivo data to determine the optimal T1-time threshold to differentiate malignant/non-malignant fluids. Inter-scan and intra-scan reproducibility was assessed by the intra-class correlation coefficient (ICC).
Ex-vivo evaluation. Patient characteristics as well as details of fluid composition of ex-vivo analysis are given in Table 1. MOLLI T1 relaxation times of malignant-positive histology fluids ranged from 1867 to 2419 ms, in malignant-negative histology from 1999 to 3223 ms and in non-malignant-negative histology from 2489 to 2897 ms; one-way ANOVA for intergroup comparison: F(2.45) = 36.1; p < 0.001; post hoc test for all intergroup comparisons: < 0.05; Fig. 3. Also, MOLLI T1 relaxation times were shorter when comparing histologically proven malignant fluids (2237.1 ± 136.9 ms) and all other fluids (2583.1 ± 249.2 ms, p < 0.001).
Linear regression established an association between protein content and MOLLI T1-measurements in our ex-vivo setup.

Discussion
Although multiple noninvasive visualization techniques have been suggested for the differentiation of malignant and non-malignant effusions 10,11 , until now no non-invasive tool has been clinically implemented. As a direct visualization of tumor cells is not possible with imaging techniques currently available, a reliable surrogate marker is necessary to indicate whether an effusion is malignant or not. Apart from cancerous and inflammatory cells, malignant effusions typically include protein 12 . Protein content is a significant determinant of T1 relaxation time, especially in fluids where relaxation times are otherwise very long 13 . T1 Mapping has previously been investigated in earlier studies dating back to the 80 s and 90 s 6,7 . However, ordinary T1 Mapping is extremely time-consuming, especially when a large field of view is employed.   www.nature.com/scientificreports/ MOLLI T1 Mapping could therefore potentially serve as a clinical applicable tool for the evaluation of effusions, as acquisition is rapid and results remain robust 9 . Until now MOLLI has only been validated for cardiac and hepatic T1 Mapping, thus we initially performed phantom analysis in fluid solutions where we found good agreement with a validated reference for T1 Mapping 14 and strong correlation with albumin content (r = − 0.97, p < 0.01).  and protein content (independent variable). Linear regression established that for ex-vivo measurements, protein content was associated with MOLLI T1-relaxation time measurements (r = − 0.69) Effusions were defined as non-malignant/negative histology (green squares) if patients had no known malignancy and histology/cytology was negative; as malignant/negative histology (red star) if patients had known malignancy but histology/cytology was negative; as malignant/positive histology (yellow circle) if patients had known malignancy and histology/cytology was positive. www.nature.com/scientificreports/ As a next step, we investigated the diagnostic significance of rapid T1-Mapping for the differentiation between malignant and non-malignant effusions. Ex-vivo regression analysis showed that apart from protein content, none of the other investigated factors were associated with T1-mapping results. These results are in line with previous data that suggest T1-relaxation rates (1/T1) of ascites are linearly proportional to protein content in non-malignant cases 7 . In contrast to non-malignant ascites, where paramagnetic ions are only marginally present 15 , higher concentrations of iron can be found in malignant ascites due to secondary iron overload. As the paramagnetic contribution of iron increases relaxation 7 , this phenomenon is expected to increase the ability to separate malignant and non-malignant ascites by means of T1-mapping.
As shown in Figs. 3 and 4, T1-relaxation times differed considerably in patients with proven malignant effusion and patients without malignancy, while results partially overlapped in patients with malignancy/positive histology (in effusion) and patients with malignancy/negative histology (in effusion). This may be explained by the composition of the various effusions. Albumin content as well as cell count were higher in effusions from patients with malignancy/negative histology than in patients without malignancy. The reason for the discrepancy between compositions remains unclear. Patients from the malignancy/negative histology group may have been false negatives in histologic assessment, as sensitivity varies greatly among different tumor types. Routine cytological evaluation of pleural fluid has a diagnostic yield ranging from 62 to 90% in patient populations with malignant pleural effusions 16 . As oncologic patients are prone to infectious complications, another likely explanation may be non-malignant effusions secondary to infections only indirectly linked to the respective tumor. These issues may be sorted out in future analyses when utilizing methods beyond cytology to invasively differentiate malignant from non-malignant effusions (i.e. GLUT-1, tumor markers etc).
Unlike TMIX, MOLLI may be used for T1-Mapping in non-stationary tissue (e.g. heart and liver), due to the much shorter acquisition duration of MOLLI T1-Mapping compared the reference sequence. This is particularly relevant, as robust mapping of pleural effusions and ascites requires fast acquisition in order to prevent breath excursions or bowel movements from falsifying results.
There are several limitations of this study. This study was conceived as a feasibility study: our patient cohort (48 study participants) was relatively small and all examinations were performed ex-vivo at 21 °C. T1-mapping has been shown to be temperature dependent 7 ; thus T1 cut-off values are expected to vary in-vivo. T1 relaxation times are reduced in fluids with elevated protein content, thus apart from malignant effusions, other exudative effusions (e.g. bacterial peritonitis, pleural empyema) will typically also show increased protein content. Although infectious effusions were not included in the current study, it is to be expected these will also show reduced T1 relaxation times. Thus, in cases where an infectious cause of effusions cannot be excluded clinically, T1-mapping may be of reduced benefit. This is the first study investigating the value of rapid non-invasive T1-Mapping for www.nature.com/scientificreports/ the differentiation of malignant/non-malignant ascites and pleural effusions. With regard to abdominal imaging, differentiating peritoneal carcinosis from non-malignant (exsudative) diseases with increased protein concentrations such as bacterial peritonitis may remain difficult. The impact of paramagnetic ions in malignant ascites on T1-Mapping results was not additionally investigated in this study, however it has previously been shown that paramagnetic ions lead to more extensive T1-relaxation time reduction than may be expected based on protein content 7 .
Overall, this promising preliminary data warrants further investigation within a larger cohort and in an invivo context in order to answer the remaining questions comprehensivelyThe results presented here indicate that MOLLI T1-Mapping may serve as a rapid non-invasive diagnostic tool for the discrimination of malignant ascites and pleural effusions. Hopefully, in the future it will help guide clinicians in this often diagnostically difficult field.