The Use of Chest Magnetic Resonance Imaging in Malignant Pleural Mesothelioma Diagnosis

In recent years, many articles have demonstrated that magnetic resonance imaging (MRI) may be performed successfully in the study of the chest. The aim of this study was to evaluate the potential role of MRI in the differentiation of benign from malignant pleural disease with a special focus on malignant pleural mesothelioma and on MRI protocols. A systematic literature search was performed to find original articles about chest MRI in patients with either benign or malignant pleural disease. We retrieved 1246 papers and 17 studies were finally identified as being in accordance with our purpose. For a morphologic assessment, T1-weighted and T2-weighted sequences were usually performed, eventually associated with T1 post-contrast sequences for better detection of pleural lesions. Functional sequences such as Diffusion Weighting Imaging (DWI), associated with the evaluation of Apparent Diffusion Coefficient (ADC) maps, were lately and gradually introduced in chest MRI protocols and their potentiality in differentiating benign from malignant disease has been investigated by many authors. Many progresses have been performed to improve quality images and diagnostic performances of MRI. A better and early identification of pleural disease may be obtained, providing MRI as a possible tool that can differentiate malignant from benign pleural disease without using invasive procedures.


Introduction
Malignant Pleural Mesothelioma (MPM) is a rare aggressive tumor that derives from mesothelial cells found in the pleural and peritoneal surface [1]. MPM is the most common primary pleural neoplasm, although metastases are still the most frequent pleural malignant lesions [2]. MPM is strongly related to asbestos exposure in 40-80% of patients, with a latency period of 45 years [2]. In addition, previous chest radiotherapies (for lymphoma, breast or lung cancers), simian virus 40 infection (SV40) and oncogenic mutations such as the Breast Cancer gene 1-associated protein (BAP 1) were highlighted as potential risk factors for MPM [2][3][4]. Male patients are most affected (men to women ratio of 4:1) with a median age among 50 and 70 years [2,5].
Patients with MPM are often symptomatic with dyspnea and chest pain; other symptoms may be cough, malaise and weight loss [2]. The presence of monolateral pleural effusion is a frequent sign (79%). Mediastinal lymph nodes and involvement of surrounding structures may cause dysphagia, superior vena cava syndrome, phrenic nerve paralysis, and cardiac tamponade [2,5].
MPM has a poor prognosis with a median survival of 4 to 18 months [2]. Nowadays treatment options include chemotherapy, targeted therapy and immunotherapy, radiotherapy and surgery [2,5,6]. Poor performance status, male gender, non-epithelioid subtype,

Materials and Methods
A systematic literature research was performed to identify all relevant data on the potential role of MRI in differentiating benign from malignant pleural disease.

Search Strategies
The databases utilized for the research of relevant publications were Scopus, Web of Science and PubMed. The last search was run on 20 November 2021. The keywords used to identify articles included benign pleural disease; malignant pleural disease; malignant pleural mesothelioma; magnetic resonance imaging; MRI. Filters were applied to include only articles published in English, described as original research and published after the year 2000. Only articles with a sample size greater than 10 patients were included.
Two authors (AM and FV) independently performed a first selection of the articles based on the title. Subsequently, the abstracts of the identified studies were screened and the full text of studies that passed the title and abstract screening was read. Any disagreement was overcome by discussion and reaching a mutual agreement.

Results
A total of 1246 studies were obtained from the research conducted. In total, 1090 records were removed for lack of inherence, while there were 112 duplicates. Titles and abstracts of the remaining 44 records were screened and after this process 35 articles were assessed for eligibility. Each article was fully read and after this process 18 papers were excluded because they did not meet all eligibility criteria. Finally, 17 studies were included in this review (Figure 1), (Table 1). enhanced (DCE) acquisition. This could be important for lesion characterization and tumor response evaluation after treatment [8].
The aim of this study was to evaluate the potential role of MRI in the differentiation of benign from malignant pleural disease with a special focus on malignant pleural mesothelioma and on MRI protocols.

Materials and Methods
A systematic literature research was performed to identify all relevant data on the potential role of MRI in differentiating benign from malignant pleural disease.

Search Strategies
The databases utilized for the research of relevant publications were Scopus, Web of Science and PubMed. The last search was run on 20 November 2021. The keywords used to identify articles included benign pleural disease; malignant pleural disease; malignant pleural mesothelioma; magnetic resonance imaging; MRI. Filters were applied to include only articles published in English, described as original research and published after the year 2000. Only articles with a sample size greater than 10 patients were included.
Two authors (AM and FV) independently performed a first selection of the articles based on the title. Subsequently, the abstracts of the identified studies were screened and the full text of studies that passed the title and abstract screening was read. Any disagreement was overcome by discussion and reaching a mutual agreement.

Results
A total of 1246 studies were obtained from the research conducted. In total, 1090 records were removed for lack of inherence, while there were 112 duplicates. Titles and abstracts of the remaining 44 records were screened and after this process 35 articles were assessed for eligibility. Each article was fully read and after this process 18 papers were excluded because they did not meet all eligibility criteria. Finally, 17 studies were included in this review (Figure 1), (Table 1).

Morphological Evaluation
When evaluating pleural lesions on MRI, a morphological assessment should be performed primarily. To achieve proper anatomical imaging and to reduce susceptibility artifacts, the Turbo Spin Echo (TSE) T2w sequence with a short echo spacing sequence, on axial, coronal and sagittal plans is recommended [9,15,17,22]. TSE T2w sequences with fat suppression are also suggested [15]. To reduce motion artifacts, breath-hold acquisition and cardiac triggering should be considered. In clinical practice many sequences are obtained with breath-hold techniques in inspiration, but the application depends on patient compliance [9]. Respiratory gating could be an alternative [16].
Malignant pleural mesothelioma shows an inhomogeneous hypointense to isointense signal on T1w and hyperintense signal on T2w MR sequences ( Figure 2) and enhancement after contrast administration [9,16]. Mediastinal pleural involvement, nodular and/or circumferential pleural thickening (more than 1 cm), nodularity and/or infiltration of adjacent structures as chest wall or diaphragm are suggestive of malignant pleural disease [16,23,24]. Typically, a retraction of the involved hemithorax is detected, also called by Hierholzer et al., "shrinking lung" [24,25]. In 2021, Sabri et al., highlighted MRI morphological features such as contour and thickness as potential tools to differentiate malignant from benign lesions with a sensitivity of 89.29%, specificity of 76%, positive predictive value of 89%, negative predictive value of 76.92% and accuracy of 85.37% [26].
In 2004, Weber et al. evaluated the role of a very short T2* gradient echo pulse sequence with fat suppression and radial K-space acquisition. The MRI protocol also included highresolution breath-hold T1w TSE, a respiratory-gated T2w TSE, half-Fourier single-shot turbo spin-echo (HASTE) sequences and finally a contrast-enhanced T1w acquisition with fat suppression. Radial MRI application provides a shorter acquisition time because of its ultrashort echo time of 0.5 milliseconds. Due to the repeated sampling of the k-space, it is helpful in reducing the effect of motion artifact due to cardiac and breath activity [16].
In a study by Plathow et al., HASTE sequences were defined as optimal for tumor delineation and pleural fluid evaluation [19]. In this study, in addition to the mentioned morphological features, volumetric analysis was proposed to assess tumor and lung volumes. For this purpose, first an automatic segmentation was attempted, then a semiautomated segmentation was preferred as the automatic one was affected by signal intensity inhomogeneities and motion artifacts. Patients also underwent spirometry. The volumetric evaluation showed a good correlation with treatment response and spirometry [19].

Functional Imaging
CT is known to be a feasible and fast imaging modality for the assessment of structural changes in many chest pathologies, including pleural disease. The development of new MRI techniques, such as functional ones, has also led MRI to gain more importance in the study of the chest, where the use of contrast agent has proven to be useful for a better pleural evaluation [17,22], therapeutical planning and for the assessment of tumor response [18,19].
Diffusion weighted imaging may be useful for a better detection of pleural disease, helping in the differentiation of a benign origin of the disease from a malignant one and to distinguish malignant lesions from each other, also thanks to the involvement of postprocessing analysis (ADC maps) [27,28].
A concordance on a threshold value for mean ADC able to differentiate lesions is still missing, although different technique of region of interest (ROI) positioning has been investigated to reduce errors in the collection of these values [29,30].
An easy, fast way of recognizing of malignant pleural lesions was also assessed and proposed, as to avoid the use of ADC [25].

Dynamic MRI
In 2006, Plathow et al., combined morphological and functional evaluation comparing standard 2D MRI and novel 3D dMRI (dynamic MRI) techniques. Functional MRI is of great importance not only in surgical/therapeutical planning but even in the assessment of tumor response. Two-dimensional dMRI was performed using true-FISP (time-resolved true fast imaging with steady-state precession) sequence measuring the displacement of chest wall and diaphragm on a workstation (SYNGO, Siemens Medical Solutions, Erlangen, Germany). Three-dimensional dMRI was obtained recording lung movement during breathing cycles in three dimensions (3D) using an isotropic time-resolved 3D gradient echo pulse sequence associated with the view-sharing implementation TRICKS (time-resolved interpolated contrast kinetics). Three-dimensional dMRI was found to provide functional parameters indicative of therapeutic response not available with simple spirometry. Regardless, 3D dMRI had a low temporal resolution and a long postprocessing time compared to 2D dMRI [18].

Contrast Enhancement-MRI
For pleural lesion evaluation, intravenous contrast administration is mandatory. For this purpose, a 3D gradient echo with fat suppression sequence is suggested with an initial basal acquisition and subsequent post-contrast study [17,22]. Pleural neoplastic lesions, in particular MPM, have a high signal intensity after contrast injection, especially in the late phase. Therefore, a post-contrast study should last at least 5 min [15,17]. Three different sequences commonly used for DCE-MRI studies were evaluated in a study by Ng et al., (2020); the authors compared radial stack-of-stars DCE-MRI, the standard cartesian-based gradient-recalled echo sequence (fast low angle shot, FLASH) and the view-sharing timeresolved imaging with stochastic trajectories methods (TWIST). Radial DCE-MRI showed optimal motion robustness in thoracic imaging with a minimum decrease in signal-to-noise ratio (SNR) compared to FLASH and TWIST sequences [17]. To differentiate between malignant and benign pleural lesions, a novel, semi-objective biomarker, the early contrast enhancement (ECE), was proposed in a study by Tsim et al. [23]. Sixty-six patients with suspect MPM were enrolled in the study. After contrast, injection images were acquired at 40 s, 80 s, 4.5 min, 9 min and 13.5 min. For each different lesion, 5 (in macronodular disease) to 15 (in non-nodular disease) ROIs were identified. The presence of lesion enhancement within 4.5 min was defined by the authors as ECE. Lesions without ECE were defined as benign. Lesions with malignant morphological features were confirmed as malignant despite the presence/absence of ECE, while lesions with benign morphological features associated with ECE were also classified as malignant. ECE evaluation as a potential biomarker showed high inter-observer agreement, with good sensitivity and negative predictive value compared to the simple morphological evaluation provided by CT and MRI without perfusion analysis [23].
Tumor response assessment is another critical issue. Modified Response Evaluation Criteria in Solid Tumors (mRECIST) are usually applied, measuring lesion thickness. CT is still the first line imaging method for MPM mRECIST application [19], but MRI could be more sensitive for this purpose. In 2008, Plathow et al. compared the application of conventional RECIST and mRECIST criteria on both MRI and CT exams, assessing that mRECIST criteria application on MRI could help in the evaluation of early therapeutic response [19] (Figure 3).  The evaluation of tumor vascularization, especially with the development of antiangiogenic drugs, has an important prognostic role. DCE analysis may provide information of lesion response to therapy when morphological evaluation is failing. A quantitative analysis of tumor enhancement was proposed by Tomšič et al., in 2019. DCE parameters were evaluated using two different kinetic models: the extended Tofts model and the adiabatic approximation of tissue homogeneity model [15].

Diffusion-Weighted Imaging and Apparent Diffusion Coefficient
DWI is a type of functional imaging based on the random movements of water molecules in tissues. Many pathologies, such as malignancies and inflammatory disease, can alter water diffusion causing a restriction of molecules movements because of molecular and structural changes [32]. Through post-processing algorithms, it is possible to represent a quantitative analysis of data obtained through DWI as apparent diffusion coeffi- The evaluation of tumor vascularization, especially with the development of antiangiogenic drugs, has an important prognostic role. DCE analysis may provide information of lesion response to therapy when morphological evaluation is failing. A quantitative analysis of tumor enhancement was proposed by Tomšič et al., in 2019. DCE parameters were evaluated using two different kinetic models: the extended Tofts model and the adiabatic approximation of tissue homogeneity model [15].

Diffusion-Weighted Imaging and Apparent Diffusion Coefficient
DWI is a type of functional imaging based on the random movements of water molecules in tissues. Many pathologies, such as malignancies and inflammatory disease, can alter water diffusion causing a restriction of molecules movements because of molecular and structural changes [32]. Through post-processing algorithms, it is possible to represent a quantitative analysis of data obtained through DWI as apparent diffusion coefficient (ADC) maps [32]. DWI and ADC maps are currently performed during MRI examinations in many oncological settings, including chest evaluation and assessment [33].

ADC Cut-Off
Many studies have been conducted to confirm that both DWI and ADC maps can be used in the evaluation of MPM or malignant pleural disease (MPD) and in differentiating benign from malignant disease. Moreover, a mean ADC value that could reliably be used for distinguishing malignant from benign lesions has been investigated too.
In 2012, Coolen et al. prospectively evaluated the use of DWI in the differentiation of MPD from benign lesions [22]. With a 3T whole-body system, DWI was performed with multiple b values (0, 50, 100, 500, 750 and 1000 s/mm 2 ), followed by DCE imaging acquisitions, used retrospectively for the evaluation of misclassified lesions. A statistically significant difference between ADC values of MPD and that of benign lesion was obtained; a mean ADC value of 1.52 × 10 −3 mm 2 /s was suggested as an optimal cut off (71.4% sensitivity, 100% specificity, 87.1% accuracy, 81% NPV and 100% PPV). Using an ADC range between 1.52 and 2.00 × 10 −3 mm 2 /s a misclassification of lesions may occur; the authors hypothesized as the presence of necrosis or inflammation in the tumor as a probable explanation. The use of DCE in addition to information gained through ADCs improved the sensitivity to 92.8% with, regardless, a decrease in specificity to 94.1%. Accuracy, PPV and NPV all changed to 93.5%, 92.8% and 94.1%, respectively [22].
In accordance with these results are those ofİnan et al., (2016); in this work the authors tried to differentiate metastatic malignant lesions from benign pleural thickening through DWI. Multiple b values were used to acquire diffusion-weighted images (0, 650 and 1000 s/mm 2 ) and multiple ADC maps were obtained (ADC 1 calculated from b value of 0 and 650 s/mm 2 , ADC 2 obtained from 0 and 1000 s/mm 2 ). A statistically significant difference of ADC 1 and ADC 2 between metastatic pleural thickening and benign lesions was found (p < 0.05). The authors found that a mean ADC value ≤ 1.5 × 10 −3 mm 2 /s suggests a malignant etiology, despite low sensitivity and specificity [27].
More recently, Usuda et al., in 2019, evaluated DWI and ADC maps to differentiate MPM from other pleural pathologies such as metastases from lung cancer, empyema and pleural effusions [31]. DWI was performed with b values of 0 and 800 s/mm 2 through a 1.5T superconducting magnetic scanner after anatomical T1w and T2w acquisitions. The mean ADC value obtained for MPM was not significantly different from that of pleural dissemination, whereas a significant difference was found between the ADC values of MPM and those of empyema and pleural effusion, and the same was reported for metastatic lesions. Both MPM and metastatic lesions presented a significantly lower ADC value than that of pleural effusion and empyema [31].
The results obtained by Sabri et al., in 2021 are similar [26]. A 1.5T scanner was used to acquire diffusion-weighted images with three different b values of 0, 500 and 1000 s/mm 2 . A qualitative assessment of the obtained images and a quantitative one with ADC maps were made. All lesions that presented diffusion restriction were proven to be histopathological malignant; conversely, all non-restricted lesions appeared to be benign. The mean ADC value of MPD was significantly lower than that of benign lesions, while no significant difference was found between the ADC value of MPM compared to that of metastatic pleural lesions, as in accordance with Usuda et al. [26,31]. Authors suggested a mean ADC cutoff value of 1.68 × 10 −3 mm 2 /s as capable of differentiating malignant from benign pleural lesions (p < 0.001) [26].
A different result was reported by Jiang et al. in 2021; retrospectively analyzing the diagnostic performance of DWI using a 1.5T superconducting magnet in patients suspected of having pleural malignancies, the author found out that although hyperintense pleural areas on DWI are suggestive of malignancy, especially with high b values, the mean ADC value in the group of pleural malignancies was not statistically different from that of the benign group, nor they could find a cutoff value that was able to discriminate malignant lesions from benign ones. The authors supposed that these results may be the consequence of the impossibility to obtain the ADC value in a large group of benign lesions and that a subjective position of the ROIs chosen for ADC evaluation may have led to bias [30].
Thus, a mean ADC value able to discriminate reliably MPD from benign lesions is still debated and more studies are needed to assess a robust cutoff value ( Table 2).

ADC-ROIs
The problem of different ROIs positioning in the ADC measurements of pleural abnormalities related to intra-and interobserver variability was assessed by Priola et al. in 2017 [29]. Five different methods of measurement were proposed. Three were based on a manual position of the ROI comprehensive of the entire circumference: Whole Tumor Volume, Three Slice Observer Defined and Single Slice; the other two consisted of the positioning of one (One Small Round ROI) or more than one (multiple small round ROI) small circular ROI in the more restricted area. Each method presented a good to excellent intra-and interobserver concordance, but better results were obtained when the entire tumor in one or more slices was considered, especially with the Single Slice method [29].

ADC-Histologic Subtype
The capability to differentiate histologic subtypes of MPM without using an invasive approach was proposed by Gill et al., in 2010 [28].
Through a 3T magnetic resonance, free-breathing DWI was acquired with different b values (250, 500, and 750 s/mm 2 ). The final group consisted of 50 investigated patients affected with histologically proven MPM (35 epithelioid, 10 biphasic and 5 sarcomatoid). Each different histologic subtype showed different average ADC values: for the epithelioid it was 1.31 ± 0.15 × 10 −3 mm 2 /s, for the biphasic it was 1.01 ± 0.11 × 10 −3 mm 2 /s, and for the sarcomatoid it was 0.99 ± 0.07 × 10 −3 mm 2 /s. No statistically significant difference was found between ADC values of biphasic and sarcomatoid tumors, while the ADC of epithelioid MPM appeared to be significantly higher than those of biphasic and sarcomatoid MPM. ADC value of the epithelioid subtype showed a sensitivity of 60%, a specificity of 94% and an accuracy of 84% in the differentiation from the sarcomatoid subtype [28].

DWI-Visual Assessment
For a more clinical and rapid assessment of pleural lesions, in 2014, Coolen et al., suggested a new visual evaluation of MPD through DWI in a group of patients suspected of being affected by malignant pleural mesothelioma [25]. The analysis of multiple b values (0, 50, 100, 500, 750 and 1000 s/mm 2 ) led them to coin the term "pleural pointillism" to report the appearance of multiple hyperintense spots mostly at high values of b parameter (1000 s/mm 2 ) as result of MPD in visual imaging examination. This diagnostic marker presented higher values of sensitivity, specificity, PPV, NPV and accuracy (92.5%, 78.8%, 89.9%, 83.9% and 88%, respectively) compared to those obtained at the CT evaluation of mediastinal thickening and shrinking lung appearance. The authors suggested that this pattern of visualization may help in pleural biopsy and limit samples [25].
In accordance with these results are those reported in 2021 by Jiang et al. The authors found areas of signal hyperintensity, with b values of 800 s/mm 2 and resembling those of pleural pointillism, in 94% of patients with malignant pleural lesions [30].
A different type of appearance of MPM in DWI was reported by Usuda et al. [31]. The authors subclassified different patterns of diffusivity restriction observed in pleural assessment into four categories (strong continuous diffusion, strong scattered diffusion, weak continuous diffusion, and no decreased diffusion). All MPM evaluated presented a strong continuous diffusion pattern on DWI, while a strong scattered pattern was observed mainly in pleural dissemination. No decrease in diffusion was observed in pleural effusion, while the weak continuous pattern was characteristic of empyema [31].

MRI Protocol: A Proposal
On the basis of the papers found in literature, an MR protocol for patients with MPM was hypothesized and elaborated. The protocol was developed using a 1.5T MRI system (MAGNETOM Symphony, Tim System, Siemens). An eight-channel phased array coil was used, and the patient was in supine position.
We elaborated an MRI scan protocol with pre-contrast acquisitions, including functional scans (DWI) and post-contrast acquisitions, performed at different acquisition times. We used commercially available sequences, while experimental MRI techniques were not considered, letting their use remain limited to a research setting. The proposed MRI protocol for MPM patients is summarized in Table 3.  For a morphological analysis, we proposed a 2D T2-weighted acquisition (Half-Fourier acquisition single-shot turbo spin-echo, HASTE, with and without fat suppression) on an axial, coronal and sagittal plan. Sequences were acquired at end expiration during multiple short breath-hold. A T1-weighted turbo spin echo (TSE) sequence was made to complete the morphological assessment.
Post-processing analysis of the data was also performed with a commercially available software, generating ADC maps.

Conclusions
MRI application for chest disease evaluation has long been discussed. In this study we revised the potential role of MRI in the assessment of MPD and in the differential diagnosis between malignant and benign lesions. An MRI protocol in order to evaluate MPD Finally, T1-weighted fat-saturated 3D gradient echo (VIBE) sequences before and after contrast injection were made. After the administration of contrast injection (0.2 mL/kg of gadoteridol (Prohance ® ; Bracco, Milan, Italy), acquisitions occurred at 40 s, 80 s, 3 min, 4:30 min and at 5 min, to assess lesions enhancement.
Post-processing analysis of the data was also performed with a commercially available software, generating ADC maps.

Conclusions
MRI application for chest disease evaluation has long been discussed. In this study we revised the potential role of MRI in the assessment of MPD and in the differential diagnosis between malignant and benign lesions. An MRI protocol in order to evaluate MPD with a special focus on MPM was proposed.
Further studies are needed to confirm and improve MRI application for pleural and general chest disease assessment.