Malignant pleural mesothelioma (MPM) is an uncommon and highly lethal neoplasm originating from the mesothelial cells lining the visceral and parietal pleura. MPM is correlated with asbestos exposure [1–3]. The incidence is increasing worldwide, and 2,000–3,000 new cases are diagnosed annually in the United States [4, 5]. MPM is generally divided into three histologic subtypes: epithelioid, sarcomatoid, and biphasic. There is a significant difference in prognosis between epithelioid and nonepithelioid (biphasic and sarcomatoid) MPM [6], the epithelioid type having a better prognosis than the other two. Because of this difference, information on histologic subtype is important as a prognostic indicator.

MPM is extremely difficult to manage regardless of the stage, histologic subtype, and treatment options. The median survival period is 9–17 months from the time of diagnosis [7]. Cytoreductive extrapleural pneumonectomy combined with chemotherapy and radiation therapy is the therapy of choice for resectable disease. Unresectable disease can be managed with a combination chemotherapy regimen that includes platinum and antifolate agents [8]. Because of the scarcity of treatment options, it is vital to develop a noninvasive method that can be used both to map the in vivo pathophysiologic characteristics of MPM and to assess the biologic response to therapy.

Several imaging techniques, including CT, MRI, and FDG PET/CT, have been used for staging of MPM and for treatment planning [9–11]. Diffusion-weighted MRI (DWI) has the potential to reveal tissue characteristics based on the diffusivity of water molecules within the tissues. With this technique, signal loss can be quantitatively assessed with the apparent diffusion coefficient (ADC), which depends on restriction of water molecule diffusion by cell membranes and macromolecules, indirectly providing information about tissue cellularity [12, 13]. DWI has been widely used in the evaluation of CNS disorders, including acute cerebral infarction, intracranial tumors, and demyelinating disease [14], and ADC has been used to characterize glioma [15]. DWI is being used to evaluate multiple intrathoracic and intraabdominal organs [16, 17]. Matoba et al. [18] correlated the ADC with histologic subtypes of lung cancer and tumor cellularity.

To our knowledge, DWI has not been previously used to evaluate MPM, and we have found no published data correlating DWI-derived ADC values with MPM subtypes. The purpose of this study was to perform DWI in the evaluation of MPM by using a free-breathing single-shot spin-echo echo-planar imaging (EPI) sequence to assess ADC and to correlate the values with the three histologic subtypes of MPM.

Sixty-two patients (51 men, 11 women; mean age, 64 years; range, 30–91 years) with known pleural abnormality and suspected MPM were enrolled in this study between June 2008 and January 2009. The institutional review board approved the study, and written informed consent was provided by all subjects.

All 62 patients underwent surgical procedures, and the histologic specimens of the lesions were analyzed for diagnosis. The procedures included extrapleural pneumonectomy (n = 16), pleurectomy (n = 7), pleural biopsy (n = 38), and cytologic examination of pleural effusion obtained with thoracentesis (n = 1).

Five of 62 suspected cases of MPM initially diagnosed clinically and with CT were histologically not MPM. The diagnoses in these cases included two benign pleural plaques, one case of nonspecific chronic inflammation, one metastatic pleural tumor, and one malignant lymphoma. The ADC could not be calculated in seven of the 57 MPM cases because the tumor volume was too small for visualization of the lesions or for reliable measurement of ADC (n = 5; three epithelioid, one biphasic, and one sarcomatoid MPM); the ADC map had severe image degradation (one epithelioid MPM); and the tumor had severe necrosis (one epithelioid MPM). The 50 MPMs with measured ADC values included 35 epithelioid, 10 biphasic, and five sarcomatoid MPMs (Appendix 1).

The average ADC in each histologic subtype of MPM was 1.31 ± 0.15 (SD) × 10^–3 mm²/s for epithelioid, 1.01 ± 0.11 × 10^–3 mm²/s for biphasic, and 0.99 ± 0.07 × 10^–3 mm²/s for sarcomatoid tumors. The distribution of ADC values by histologic subtype is shown in Figure 4. The ADC of the epithelioid subtype was statistically significantly higher than the ADC values of the biphasic and sarcomatoid subtypes (p = 2.48 × 10^–4, p = 8.14 × 10^–5, respectively). There was minimum overlap in the ADC distribution of epithelioid and sarcomatoid subtypes, but 32 of 35 cases (91%) of epithelioid tumors had higher ADC values than did sarcomatoid tumors (cutoff ADC threshold, 1.1 × 10^–3 mm²/s). The sensitivity, specificity, and accuracy for differentiating epithelioid and sarcomatoid MPM on the basis of ADC value at a threshold of 1.1 × 10^–3 mm²/s were 60%, 94%, and 84%. The ADC values of biphasic MPM had a wide range of overlap with the ADC values of other subtypes, ranging from 0.9 × 10^–3 mm²/s to the highest value's being closer to the mean epithelioid ADC of 1.31 ± 0.15 × 10^–3 mm²/s, even though there was a statistically significant difference across the three types (F_2,55 = 7.61, p = 0.001, analysis of variance). There was no statistically significant difference between the ADC values of biphasic and sarcomatoid MPM (p = 0.095).

In the two cases of benign plaque, the ADC values were similar to the ADC of adjacent muscle and lower than that of all the histologic MPM subtypes. The average ADC in the two cases of benign plaque was 0.85 ± 0.17 × 10^–3 mm²/s.

Differences in water molecule diffusivity within different tissue types can be detected with DWI [22]. High-grade gliomas such as glioblastoma tend to have lower ADC values than do low-grade gliomas such as astrocytoma. High-grade gliomas have higher cellularity and expanded extracellular space with greater tortuosity and a more sophisticated histologic structure than do low-grade gliomas [23, 24]. Such factors restrict water molecule motion and consequently lower the ADC of high-grade gliomas. Similar results have been reported for other tumors, including lung cancer [25].

Sarcomatoid MPM is composed of spindle-shaped cells with relatively high cellularity, and epithelioid MPM is composed of tubular, cuboidal cells. Sarcomatoid MPM often forms a storiform pattern, which can complicate the histologic structure and further restrict water diffusion. In addition, sarcomatoid MPM, with its higher microvascular density, micronecrosis, and cell edema, tends to undergo more aggressive tumor growth than the epithelioid subtype and has a large nucleus-to-cell ratio. These factors can contribute to the much lower ADC of sarcomatoid than of epithelioid MPM.

The ADC values of biphasic MPM had a wide range of overlap with the ADC values of other subtypes. Biphasic MPM consists of a mixture of both epithelioid and sarcomatoid cells with assorted volumes of high cellularity and low cellularity. It therefore is reasonable for this subtype to have moderately high average ADC values overlapping with the values for both epithelioid and sarcomatoid MPM. Furthermore, the ADC values were based on an average of three ROIs within a tumor and therefore depended on the location of the ROIs. The value thus may have varied depending on whether an ROI was located within more epithelioid or more sarcomatoid tissue.

The histologic subtype of MPM is usually determined at pleural cytologic examination, needle biopsy, pleurectomy, or examination of a specimen obtained by video-assisted thoracoscopy. The sensitivity of pleural cytologic evaluation is only 20–33% [26]. Although the diagnostic accuracy of video-assisted thoracoscopy is high (greater than 95%), the procedure is invasive and may lead to seeding of tumors along the surgical incision or chest tube track in as many as 20% of patients [9]. Compared with these methods, DWI is noninvasive and can be complementary to MRI with other sequences routinely used to assess resectability and establish a treatment plan. DWI may have great potential in the noninvasive diagnosis of subtypes of MPM.

In the two cases of benign plaque, in which the ADC values were similar to those of adjacent muscle and were lower than those of all of the MPM subtypes, it may be that the cellular structure in the benign plaque was similar to the structure of normal tissue determined with the ADC values. However, because our cohort of patients had only two cases of benign plaque, statistical analysis was not applicable. It would be beneficial to rule out benign pleural disease by adding ADC information before invasive biopsy is considered. It is necessary to continue to investigate larger patient populations with benign plaques; there may be great potential for differentiating benign plaques from MPM on the basis of ADC.

One of the challenges in applying DWI to the thoracic region is image distortion due to respiratory motion, heart motion, and susceptibility artifacts from air–tumor interfaces. We used the GRAPPA technique [21], in which autocalibration is used to limit susceptibility in the lung, resulting in robust image reconstruction. The DW images and ADC maps obtained were satisfactory for assessment of signal intensity and for ADC measurement. Comparative studies [26, 27] with free-breathing techniques and a larger number of signals acquired, breath-holding technique, and respiratory and cardiac triggering have shown that free-breathing technique is robust not only for obtaining DW images with improved signal-to-noise and contrast-to-noise ratios but also for accurately evaluating ADC, for which more than two b values are acquired [28]. Kwee et al. [29] reported that free-breathing technique has better reproducibility in calculating ADC than does respiratory triggering technique and is as good as breath-holding techniques in imaging of the liver. Studies of DWI of the lung have been performed with cardiac gating, respiratory triggering, and breath-holding techniques [27]. Because many patients in this cohort had marked restrictive physiologic conditions due to the circumferential encasing pattern of the disease and were not capable of performing a breath-hold, we adopted free-breathing spin-echo EPI DWI. It is true, however, that free-breathing technique is subject to error in ADC calculation due to image misregistration. The error can vary with the magnitude of respiratory motion and the size of the lesion. In other words, ADC can become less accurate in patients with severe dyspnea and those with small lesions. Motion-induced error may account in part for the variability of ADC measurement in our results.

Our study had several limitations. First, the location of ROIs on the ADC maps could not be matched with the biopsy locations. Therefore, we averaged three ROIs from carefully selected nonnecrotic regions of the tumor. Especially in the biphasic subtype of MPM, the ADC can vary depending on the location of the ROIs because the tumor cell distribution is known to be inhomogeneous and the ADC value may have not reflected the histologic features. Second, the number of patients with the biphasic subtype was relatively small. Continued investigation of this subtype is necessary.

On the basis of the results of this preliminary study, we conclude that the ADC of the epithelioid subtype of MPM is higher than that of the sarcomatoid subtype, a difference that may serve as a surrogate imaging biomarker.

R. R. Gill and S. Umeoka contributed equally to this study.

Address correspondence to R. R. Gill. ([email protected]).

This is a Web exclusive article.