Radiologic Assessment of Mesothelioma

Radiologic Assessment of Mesothelioma

Imaging plays an essential role in the diagnosis, staging, and clinical management of patients with mesothelioma. X-ray imaging techniques [chest radiography and computed tomography (CT)], magnetic reso- nance imaging (MRI), and positron emission tomography (PET) have all been used to evaluate this disease, although the relative importance of these imaging modalities has changed over time. Our understand- ing of mesothelioma has been advanced through radiologic examina- tion, and nearly every mesothelioma patient makes numerous trips to the radiology department during the course of treatment. Imaging studies define the morphology and extent of mesothelioma, tumor perfusion, tumor physiology, the presence of mediastinal or chest wall involvement, and the presence of concomitant disease. The image acquisition device (i.e., the hardware) is only one component of the radiologic examination; software tools for the subsequent visualization and postprocessing of the acquired image data are required to extract useful information from the image pixels and to fully exploit the wealth of information contained within the image. This chapter describes the imaging modalities that have been employed for the evaluation of mesothelioma and emphasizes the role of CT in the important task of tumor thickness measurement for the assessment of tumor progression or response to therapy.

Imaging Modalities

Radiography

Chest radiography continues to rank as the most common radiologic procedure performed in the United States. Consequently, initial detec- tion of mesothelioma in a patient is likely to result from a radiographic chest examination. The two-dimensional radiographic projection of mesothelioma with its complex three-dimensional morphology, how- ever, provides neither a sensitive nor a specific diagnosis, and a follow-up study with another imaging modality is almost always indicated. The ability to diagnose mesothelioma on chest radiography

usually occurs at later, more advanced stages of the disease when tumor burden is greater.

Initial radiographic signs of mesothelioma include a unilateral pleural abnormality with an associated ipsilateral pleural effusion, ipsi- lateral shift of the mediastinum, and unilateral lung volume loss due to encasement of the lung by the tumor . Signs of other asbestos-related disease are usually absent, and the typical finding of diffuse lobulated pleural thickening is indistinguishable from pleural metastases (2,3). At later stages of the disease, radiography may demon- strate thickening of interlobular septa, rib or vertebral body destruction, lymph node metastases, and metastatic pulmonary nodules (4). Con- tralateral pleural abnormalities, when present, are typically the result of benign asbestos-related disease rather than metastases (5), since meso- thelioma generally spreads by contiguous growth; nevertheless, hema- togenous spread of mesothelioma may be observed on imaging studies (see Fig. 28.4) and was present in 44 of 66 autopsy cases in one series (6). Radiography plays a role in the posttherapy follow-up of patients. For example, patients who undergo extrapleural pneumonectomy may be monitored for complications and recurrence with chest radiography once the affected hemithorax has opacified (5). Findings such as medi- astinal shift, a new air–fluid level in the affected hemithorax, or nodules in the contralateral lung would indicate that a CT scan is warranted to differentiate between recurrent disease, infection, or a postsurgery complication (5) (Fig. 28.2). More often, however, CT is being used as

the sole imaging modality for routine posttherapy follow-up.

Computed Tomography

The imaging modality with the greatest impact on the current evalua- tion of mesothelioma is CT. The transaxial images generated by CT overcome the superposition of anatomic and pathologic structures that limits the two-dimensional projection images acquired by radiography. Accordingly, the spatial extent and radiologic characteristics of mesothelioma tumor may be more clearly appreciated with CT.

The radiologic manifestation of pleural response to a variety of dis- eases falls into three broad categories: pleural effusion, pleural thick- ening, and pleural calcification (7). Computed tomography is especially capable of demonstrating such pleural responses. The particular CT findings of mesothelioma, however, are not pathognomonic; a variety of benign and malignant diseases (including metastatic disease, tuber- culous pleurisy, empyema, and asbestos-related advanced pleural abnormalities) can have similar characteristics on CT (8,9).

On CT, mesothelioma is characterized by a circumferential, lobulated soft tissue mass that often involves the interlobar fissures and the medi- astinal pleura of a hemithorax (2) (Fig. 28.3); bilateral disease is rare (10). Pleural effusions (see Figs. 28.11 and 28.13A below) and nodular pleural thickening, especially in the lower thoracic zone, are typical CT findings in mesothelioma patients (5,10). A tendency for right-sided disease has been observed (10). Intravenous iodinated contrast admin- istered intravenously is typically used to identify mediastinal lymph node enlargement and to determine the relation of lesions to adjacent vascular structures (10); a recognized shortcoming of CT, however, is

its limited sensitivity for hilar lymph node involvement (10). Although pleural plaques are a common CT finding in mesothelioma, this reflects the role of asbestos exposure in the pathogenesis of both lesions; the possible preneoplastic nature of such plaques has not been proven (11,12).

In a series of 50 patients, Ng et al (13) observed that 76% of the ini- tial CT scans demonstrated pleural effusions, of which the majority were considered “small” (i.e., they occupied less than one third of the hemithorax). Pleural thickening was observed in 94% of cases, of which 72% was nodular, 50% showed a lower zone predominance, and 47% exceeded 1 cm (13). Superior mediastinal pleural thickening was ob- served in 70% of cases, diaphragmatic crural thickening was demon- strated in 84% of cases, and thickening of the pleural surfaces of the interlobar fissures was present in 84% of cases (13). Kawashima and Libshitz (14) report similar findings. In their series of CT scans from 50 mesothelioma patients, 74% of cases demonstrated pleural effusions (of which approximately half occupied less than one third of the hemitho- rax), 86% of cases demonstrated thickening of the pleural surfaces of the interlobar fissures, and pleural thickening of various extent, thick- ness, and nodularity was observed in 92% of cases. Focal pleural masses (ranging from 7 to 18 cm in maximum diameter) were observed in 8% of cases; half of these cases demonstrated chest wall invasion .

The volumetric extent of disease may be more clearly appreciated with CT than with chest radiography. The CT findings depicting the impact of mesothelioma on the affected hemithorax volume are varied. In response to volume loss of the ipsilateral hemithorax, for example, ipsilateral mediastinal shift may occur (Fig. 28.4). Alternatively, tumor encasement of the ipsilateral lung may result in ipsilateral volume loss without mediastinal shift (referred to as the “fixed mediastinum”). Ipsi- lateral volume loss may also be demonstrated on CT by narrowed intercostal spaces [so-called rib crowding (10)] and ipsilateral hemidi- aphragm elevation (14). Substantial pleural effusion or pleural thick- ening, however, may cause contralateral mediastinal shift with a corresponding increase in ipsilateral hemithorax volume. The CT section in  represents a hybrid of these mechanisms: ipsi- lateral volume loss with rib crowding combined with contralateral shift of the mediastinum. Ng et al (13) observed ipsilateral volume loss in 27% of cases, of which 68% demonstrated ipsilateral mediastinal shift; ipsilateral volume increase was observed in 10% of cases, of which 57% demonstrated contralateral mediastinal shift. It is interesting to note that the volume of the affected hemithorax was not substantially altered in 63% of cases at initial CT (13). Kawashima and Libshitz (14) observed ipsilateral volume loss in 42% of cases, of which approxi-mately half demonstrated ipsilateral mediastinal shift; contralateral mediastinal shift (due to a large effusion or a combination of effusion and mass) was observed in 14% of cases. Neither change in hemitho- rax volume nor shift of the mediastinum were observed in 44% of cases (14). Yilmaz et al (10) also noted ipsilateral volume loss with (9% of cases) and without (22% of cases) ipsilateral mediastinal shift, con- tralateral mediastinal shift due to a large effusion or a combination of effusion and mass (26% of cases), and no change in mediastinal posi- tion or affected hemithorax volume (43% of cases).

Although primary pericardial mesothelioma is rare, pericardial inva- sion of pleural mesothelioma is demonstrated at CT by pericardial thickening with potential concomitant pericardial effusion (5) (Fig. 28.6). It should be noted, however, that distinction between mediasti- nal pleural disease alone and associated pericardial disease is difficult on CT (14). Some investigators suggest that pericardial involvement should be considered likely when involvement of the mediastinal pleura is bulky or extensive at CT (10).

CT findings are often used in the differential diagnosis of diffuse pleural disease to distinguish between benign pleural disease and mesothelioma (or other malignant pleural disease). The presence of a pleural rind, involvement of the mediastinal pleura, pleural nodular- ity, and pleural thickening in excess of 1 cm have all been associated specifically with malignant pleural disease (1) and are all well depicted on CT. Moreover, invasion of the chest wall or mediastinum  displacement or destruction of ribs or vertebral bodies transdiaphragmatic growth (Fig. 28.9), and lymph node metas- tases (Fig. 28.10) are other CT-based indicators of malignancy (1), although MRI may have advantages over CT with regard to some of these indicators. In a series of 74 patients with diffuse pleural disease, Leung et al (7) observed that among the 71 patients with pleural thick- ening on CT, four CT findings—presence of a pleural rind, nodular pleural thickening, parietal pleural thickening greater than 1 cm, and mediastinal pleural involvement—were significantly more common in patients with malignant pleural disease than in patients with benign pleural disease. The three patients without pleural thickening demon- strated unilateral pleural effusions, the sole indicator of pleural malig- nancy in these patients; thus, the authors conclude that absence of pleural thickening does not preclude a malignant diagnosis. The CT findings in mesothelioma patients were the same as the CT findings in patients with metastatic pleural disease, and the CT findings that dis- tinguished mesothelioma from benign pleural disease were essentially the same as those that distinguished malignant pleural disease from benign pleural disease (7). Pleural calcifications were observed to be indicative of a benign process, since none of the 11 mesothelioma patients in this series demonstrated pleural calcifications. Although  benign pleural disease in general may present unilaterally, unilateral pleural disease within asbestos-exposed patients was highly specific for malignant disease generally and suggestive of mesothelioma in particular (7).

Computed tomography has also been shown to differentiate between mesothelioma and other malignant pleural disease, although this task has generally been considered a more difficult radiologic challenge. In a series of 215 patients (99 with mesothelioma, 39 with metastatic pleural disease, and 77 with benign pleural disease), Metintas et al (8) used multivariate analysis to show that (1) the presence of a pleural rind, (2) mediastinal pleural involvement, and (3) pleural thickness greater than 1 cm were independent findings both for differentiating mesothelioma from metastatic pleural disease and for differentiating malignant pleural disease (i.e., mesothelioma and metastatic pleural disease) from benign pleural disease. The first two findings were also useful for the differentiation of mesothelioma from benign pleural disease. Nodular pleural thickening was common among the CT scans of mesothelioma patients, and although it was found to be an inde- pendent finding for the differentiation of mesothelioma or malignant pleural disease from benign pleural disease, nodular pleural thicken- ing could not be used to differentiate mesothelioma from metastatic pleural disease (8).

Another important aspect of CT is its ability to depict ancillary find- ings in the lungs that typically accompany mesothelioma and are asso- ciated with prior asbestos exposure. These findings include ipsilateral atelectasis [observed in 74% of cases in the 70-patient series of Ng et al (13)], rounded atelectasis [observed in 9% of cases in this series (13)], and lung nodules [observed in 11% of cases (13)]. A CT finding of com- pressive atelectasis secondary to a large pleural effusion in a mesothe- lioma patient. 

Computed tomography has become a valuable tool for biopsy guid- ance. Closed pleural needle biopsy may be used in lieu of more inva- sive procedures (e.g., thoracoscopy or thoracotomy) to obtain pleural tissue or fluid samples for histopathologic diagnosis. In the absence of CT guidance, however, the sensitivity of closed pleural needle biopsy for the diagnosis of mesothelioma has been limited due to a typically small sample size and an inability to visualize the source of the acquired sample within the patient (15,16). The addition of CT guid- ance to the biopsy procedure greatly reduces these limitations. In a series of 30 patients, Metintas et al (15) correctly diagnosed mesothe- lioma in 83% of cases by use of CT-guided closed pleural needle biopsy, a figure that represents a substantial improvement in efficiency relative to the same biopsy procedure performed without CT.

Magnetic Resonance Imaging

Magnetic resonance imaging adds substantial information to the clini- cal evaluation of mesothelioma patients, particularly with regard to resectability (due to its ability to depict local tumor extent), diagnosis, staging, surgical planning, and follow-up. Most cases of mesothelioma about the ribs and chest wall, and a substantial percent also about the pericardium and diaphragm. T1-weighted MRI may be used to iden- tify edema in the ribs, a finding consistent with tumor invasion. Mag- netic resonance imaging has an advantage over CT in its ability to image tissue planes; indeed, a clear fat plane between the inferior diaphragmatic surface and the adjacent abdominal organs, plus a smooth inferior diaphragmatic surface on MRI, is one of the most reli- able indicators of resectability. Likewise, lack of tumor invasion into the mediastinal fat is another measure of resectability better demon- strated on MRI.

One generally recognized advantage of MRI over CT has been the multiplanar capabilities inherent in the MRI acquisition process. Although the spatial resolution of CT in the imaging plane exceeds that of MRI (pixel dimensions on the order of 0.7 mm versus 1.0 mm), CT image acquisition is constrained to the axial plane; postprocessing of the axially acquired data is possible to reformat sagittal and coronal image planes, but the anisotropy of traditional CT voxels renders such reformatted images with suboptimal quality compared with the axially reconstructed images. Magnetic resonance imaging, however, allows for the acquisition of images in arbitrary planes, a powerful capability for the evaluation of mesothelioma with its platelike growth pattern and propensity for chest wall invasion, diaphragmatic involvement, and extension into the interlobar fissures. The multiplanar aspects of MRI do not suffer from the partial volume effect that is characteristic  of axial CT images near curved structures such as the lung apices or the dome of a hemidiaphragm (17).

The multiplanar advantage of MRI, however, is waning in the face of newer multidetector row CT scanners. With 16 or more rows of detectors, rapid high-resolution acquisition has become possible with isotropic voxels so that no preferred plane exists for image reconstruc- tion (18). In effect, all planes have equal resolution, and the radiologist or clinician may decide, after image acquisition, which visualization plane best meets the needs of the particular study. The diagnostic eval- uation of mesothelioma is expected to benefit tremendously from this improvement in CT technology.

Magnetic resonance imaging has a further advantage over CT with regard to the information captured. Computed tomography predomi- nantly records information about one physical characteristic of patient anatomy and pathology: attenuation coefficients. An x-ray beam gen- erated by a CT scanner traverses the patient and is attenuated to a greater or lesser extent depending on the attenuation coefficients of the tissues encountered on its way to the detector; the chemical composi- tion and physical density of the material, along with the energy spec- trum of the x-ray beam, determine the fundamental appearance of    the acquired image. The myriad pulse sequences available on MRI scanners, however, are designed to capture information about dif- ferent physiologic and molecular processes within the patient. These processes include exchange of water on and off of macromolecules and membranes, water diffusion, and blood flow. The MRI pulse sequences exploit the characteristic differences between these processes in differ- ent tissues to provide the required image contrast necessary for tissue differentiation, which may be further enhanced through administration of contrast agents [such as gadolinium–diethylenetriamine pentaacetic acid (Gd-DTPA)] that advantageously alter relaxation and local mag- netic susceptibility. In this context, prediction of mesothelioma response and degree of response to new antiangiogenic agents may be captured by MRI.

In a study of 26 paired MRI and CT scans of mesothelioma patients at various stages of disease, Knuuttila et al (17) directly compared the imaging findings of MRI and CT to identify the relative merits of each modality. They found that CT exceeded MRI in its ability to depict pleural calcifications and to detect enlarged lymph nodes with patho- logic suspicion. Neither CT nor MRI, however, could be used to accurately assess lymph node staging due to low sensitivity and low specificity. The ability to depict invasion of the chest wall, medi- astinum, and lung parenchyma was found to be equal for both modal- ities. Relative to CT, MRI more clearly indicated the spread of tumor into the interlobar fissures, the extension of tumor through the diaphragm (Fig. 28.12), and tumor invasion of ribs or vertebral bodies. MRI demonstrated an important ability to differentiate mesothelioma from the pleural fluid that usually accompanies it and often confounds the assessment of tumor burden (Fig. 28.13). The authors concluded that MRI was “better for evaluating the growth pattern and extent of

[mesothelioma] and should be more widely used, especially when eval- uating tumor resectability and in research protocols when an accurate evaluation of disease extent is essential” (17).

Other authors have noted the increased signal strength of mesothe- lioma relative to the chest wall on T2-weighted MRI (5,19,20). More- over, MRI may be used to exclude tumor invasion of the spinal canal (1). With regard to diaphragmatic effects of mesothelioma, neither chest radiography nor thoracic CT is capable of distinguishing between ele- vation and inversion of a hemidiaphragm to the extent possible with MRI, which, on coronal images, depicts the diaphragm as a distinct linear structure separating intrathoracic and intraabdominal structures (21).

A study by Knuuttila et al (22) compared the relative abilities of contrast-enhanced CT and MRI to differentiate mesothelioma from other pleural malignancies or benign pleural disease, although the imaging findings of the study were not verified surgically. In a study with 34 sets of paired CT and MRI scans, the findings of pleural fluid, pleural enhancement, focal pleural thickening, and enhancement of focal pleural thickening were observed statistically significantly more frequently in mesothelioma patients than in patients with other pleural malignancies or benign pleural disease. Focal thickening and enhance- ment of interlobar fissures occurred significantly more frequently in malignant pleural disease (mesothelioma or other malignancy) than in   benign pleural disease. Magnetic resonance imaging was able to depict abnormal enhancement of interlobar fissures better than CT, but CT better depicted pleural calcifications, although calcifications were not specific to mesothelioma, other pleural malignancy, or benign pleural disease. Compared with CT, MRI better depicted invasive tumor growth into the diaphragm, mediastinum, and chest wall, findings that were observed significantly more frequently in mesothelioma patients than in patients with other pleural malignancies, and MRI better depicted invasion of bony structures, a finding that was observed sig- nificantly more frequently in patients with other pleural malignancies than in mesothelioma patients. Neither modality was able to differen- tiate pathologic mediastinal lymph nodes (22).

The role of imaging in the staging of mesothelioma has gained inter- est in recent years. In the context of the International Mesothelioma Interest Group (IMIG) Staging System (23), Heelan et al (24) compared MRI and CT findings with surgical and pathologic staging for 65 patients who underwent one of the following procedures: extrapleural pneumonectomy, thoracotomy with partial pleural pleurectomy, tho- racotomy with biopsy, laparoscopy with biopsy, or supraclavicular lymph node biopsy. Of the anatomic sites evaluated, only two demon- strated significant differences between the diagnostic capabilities of CT and MRI, with MRI demonstrating superiority over CT: invasion of the diaphragm and invasion of the endothoracic fascia or a single chest wall focus of involvement. Other anatomic sites that were evaluated under this staging system included scattered foci of visceral pleural involvement, confluent visceral pleural tumor, invasion of lung parenchyma, mediastinal fat involvement, pericardial involvement, chest wall invasion, and ipsilateral hilar or mediastinal lymph node involvement. Overall, both imaging modalities demonstrated fairly low diagnostic accuracies. These investigators suggested that the complex growth pattern of mesothelioma along pleural and fissural surfaces combined with the anatomic contiguity of the pleural tumor and the structures it eventually invades hinders the ability of cross- sectional imaging to stage mesothelioma with greater accuracy (24).

Positron Emission Tomography

Positron emission tomography with the fluorine-18-labeled analog of 2-deoxyglucose (F-18 fluorodeoxyglucose or FDG) as a radiotracer pro- vides uniquely different information from other imaging modalities. The resulting functional images of metabolic activity have been used in oncology to differentiate malignant from benign lesions, to stage malignant disease, and to assess tumor response to therapy. The bene- fits of PET imaging in recent years have gained recognition for the eval- uation of mesothelioma. In particular, its role as an adjunct to CT and MRI for the diagnosis of mesothelioma and the identification of the extent of disease has been explored.

Positron emission tomography images may be analyzed either qual- itatively (i.e., visually) or through semiquantitative metrics, such as the standardized uptake value (SUV), which measures the ratio of decay- corrected radiotracer uptake in a region (i.e., a lesion) to the injected dose normalized for body weight. In a study based on the visual inter- pretation of PET images from 15 patients, the presence of mesothelioma was detected by PET in all 11 positive cases, and the absence of disease was confirmed in the four negative cases (25). Of the 34 lesions from these cases that were biopsied, 28 of the 29 actually positive lesions were identified on PET (the one false negative measured 0.5 mm in diameter) and four of the five actually negative lesions were confirmed on PET (the one false positive was inflammatory pleuritis). Three pat- terns of FDG uptake were noted (focal or linear, diffuse, and hetero- geneous), which corresponded to the structural findings observed at MRI or CT. Whereas PET identified all three patients with chest wall involvement, CT only provided evidence of such involvement in one of these patients; moreover, PET identified bilateral disease in three patients, while CT demonstrated bilateral involvement in only one of these patients (25).

Carretta et al (26) obtained a PET-based sensitivity of 92% for the identification of mesothelioma based on visual interpretation aug- mented by SUV values. The one false negative represented mesothe- lioma of the epithelial subtype, which tends to have low metabolic uptake (27). Since it measures tissue metabolic activity of any nature, FDG is not a specific tumor marker (27); therefore, PET is unable to dis- criminate mesothelioma from other malignant pleural disease and should not replace histologic diagnosis based on biopsy or thora- coscopy (26), although the disease activity demonstrated by PET may be used to guide biopsy site selection (27).

Using semiquantitative SUV values alone, Bénard et al (27) reported a 91% sensitivity and a 100% specificity for the differentiation of malig- nant and benign pleural disease by PET. The potential staging of the extent of mesothelioma by PET was also observed (27), although others have concluded that PET does not depict local extent of mesothelioma but is valuable for the identification of extrathoracic metastases (28). Bénard et al (29) later showed statistically significantly shorter survival times among patients in a high SUV group, concluding that patients with highly active mesothelioma on PET (i.e., more metabolically active disease and hence, a greater uptake of FDG) have a poorer prognosis. The extent to which this increased FDG uptake indicated inherent biologic characteristics of mesothelioma in these patients or simply reflected differences in tumor size remained an unanswered question.

Tumor Measurement

The notion of tumor response is fundamental in oncology. Assessment of disease progression or response to therapy is necessary for the clin- ical management of the oncology patient and critical for the evaluation of drug efficacy during clinical trials. Accordingly, the diagnostic role of imaging is replaced by a surveillance role once the presence of mesothe- lioma in a patient is confirmed. The importance of this surveillance role must not be underestimated: the radiologic assessment of patients enrolled in clinical trials for the evaluation of novel therapeutic regi- mens has gained acceptance as a surrogate for patient survival out- comes during the regulatory approval process (30). Clinical trials thus may be conducted with smaller subject populations, a benefit that reduces both time and expense. This radiologic assessment, however, necessitates quantitative tumor measurements and the standardization of tumor response criteria based on such measurements.

The issue of standardization has evolved over the years. In 1981, the World Health Organization (WHO) recommended the radiologic quan- tification of solid tumors through bidimensional measurements on imaging studies (31). These measurements represent the product of (1) the length of the longest in-plane diameter of the lesion (as represented on the section that demonstrates the greatest extent of the lesion for CT or MRI scans) and (2) the length of the longest diameter that may be constructed perpendicular to the longest in-plane diameter. Tumor response then is determined from a comparison of lesion bidimensional measurements across temporally sequential imaging studies (31).

Nearly two decades later, the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines advocated the replacement of bidimen- sional tumor measurements with unidimensional measurements, specifically on CT or MRI scans: the length of the longest axial diame- ter of the lesion (on the CT section that demonstrates the greatest extent of the lesion) (32,33). Under these guidelines, a tumor is classified as demonstrating (1) partial response, when the sum of the unidimen- sional measurements of all lesions in a follow-up CT scan represents a decrease of more than 30% from the baseline scan sum; (2) progressive disease, when the unidimensional measurement sum in the follow-up CT scan represents an increase of more than 20% from the baseline scan sum (or if new lesions develop); (3) stable disease, when the extent of measurement reduction is not great enough to qualify as partial response or the extent of measurement increase is not great enough to qualify as progressive disease; or (4) complete response, when the follow-up scan demonstrates resolution of all lesions (33). These tumor response criteria were found to be in concordance with the WHO cri- teria of 50% reduction for partial response and 25% increase for pro- gressive disease (32).

The measurement guidelines offered by WHO or RECIST, which were designed for compact tumors, are generally not as appropriate for mesothelioma with its circumferential growth pattern and often scal- loped morphology (34,35). Accordingly, alternative CT measurement protocols, adapted from RECIST, have been proposed specifically for mesothelioma. For one such protocol that is gaining recognition, between one and three unidimensional measurements of pleural thick- ness are obtained on each of three CT sections (36,37). The sum of these unidimensional measurements is used to represent tumor burden. The RECIST guidelines for tumor response classification then are applied to the summed measurements obtained from temporally sequential CT scans. The actual manner in which tumor measurement protocols are implemented raises issues of consistency and reproducibility. In studies unrelated to mesothelioma, inter- and intraobserver variability in the selection and measurement of lesions in CT scans have been reported (38–40); the circumferential morphology and axial extent of mesothe- lioma, however, further complicate the measurement of this specific tumor. Such difficulties may impair accurate evaluation of patient prognosis and hinder an accurate evaluation of clinical trials. In a recent study, Armato et al (41) articulated a three-step process for the manual measurement of mesothelioma that involves (1) selection of a limited number of CT sections in which the disease is most prominent,

(2) identification of specific locations within the selected sections that demonstrate the greatest extent of pleural thickening, and (3) the actual measurement of tumor thickness at those locations. With the first two of these steps held fixed, 95% limits of agreement for relative interob- server difference of mesothelioma tumor thickness measurements were found to span a range of 30% for a database of 22 CT scans. The inves- tigators noted the expectation of increased variability had observers been allowed to implement all three steps of the measurement process and had temporally sequential scans of the patients been evaluated as they are in actual clinical practice (41). Such variability may lead to dis- cordant tumor response classification, which may adversely affect the conduct of clinical trials.

Computed tomography provides an opportunity for computerized image analysis methods to facilitate implementation of tumor mea- surement protocols. Much progress has been made in the use of com- puters to analyze medical images, and the potential of semiautomated techniques for the measurement of tumor masses in CT has been shown (42). Armato et al (41) developed a computer interface and computer- ized techniques for the semiautomated generation of mesothelioma tumor thickness measurements. User-identified points along the chest wall or mediastinal boundary are automatically connected to the lung boundary to provide pleural thickness measurements. In a study of 22 CT scans from mesothelioma patients, the mesothelioma measure- ments generated by the semiautomated algorithms closely approxi- mated the average measurements of five human observers. Of all semiautomated tumor thickness measurements, 83% were  within  15% of the corresponding average manual measurements (41). Such computer-assisted approaches are expected to greatly enhance the utility of CT scans in the management of mesothelioma patients, to reduce data acquisition time during clinical trials, and to make the radio- logic assessment of mesothelioma more efficient and consistent.

Despite the volumetric capabilities of CT, tumor volume is not con- sidered in the present clinical evaluation of mesothelioma. Some inves- tigators have begun to explore tumor volume. Pass et al (43), for example, showed a correlation between mesothelioma tumor volume and median survival in a series of 48 patients. Furthermore, Prasad et al (44) demonstrated that measurements of metastatic tumors based on volume yield tumor response classifications that differ from those obtained based on the RECIST guidelines, so that, in general, linear measurements may not be accurate surrogates for tumor volume. The fact that volume is not considered clinically, however, is out of neces-sity, not out of need. Volume measurements are needed, but such mea- surements are exceedingly cumbersome and quite impractical to obtain through manual approaches, especially for mesothelioma; clinicians, therefore, have submitted to the more practical acquisition of a limited number of unidimensional measurements. The extent to which unidi- mensional measurements sufficiently capture the often asymmetric and nonuniform three-dimensional growth of a morphologically complex tumor such as mesothelioma is questionable, but volume measure- ments will certainly require some degree of automation. To this end, the power of the computer will be more fully realized by automated and semiautomated methods that evaluate the two- and three- dimensional characteristics of tumor area and volume.