ORIGINAL ARTICLE

FDG PET/CT patterns of treatment failure of malignantpleural mesothelioma: relationship to histologic type,treatment algorithm, and survival

Victor H. Gerbaudo & Marcelo Mamede &

Beatrice Trotman-Dickenson & Hiroto Hatabu &

David J. Sugarbaker

Received: 24 September 2010 /Accepted: 2 December 2010 /Published online: 6 January 2011# Springer-Verlag 2010

AbstractPurpose This study investigated the diagnostic performanceand prognostic value of fluorodeoxyglucose (FDG) positronemission tomography (PET)/CT in suspected malignantpleural mesothelioma (MPM) recurrence, in the context ofpatterns and intensity of FDG uptake, histologic type, andtreatment algorithm.Methods Fifty patients with MPM underwent FDG PET/CT for restaging 11±6 months after therapy. Tumor relapsewas confirmed by histopathology, and by clinical evolutionand subsequent imaging. Progression-free survival wasdefined as the time between treatment and the earliestclinical evidence of recurrence. Survival after FDG PET/CTwas defined as the time between the scan and death or lastfollow-up. Overall survival was defined as the timebetween initial treatment and death or last follow-up date.Results Treatment failure was confirmed in 42 patients (30epithelial and 12 non-epithelial MPM). Sensitivity, specificity,accuracy, negative predictive value, and positive predictive

value for FDG PET/CT were 97.6, 75, 94, 86, and 95.3%,respectively. FDG PET/CT evidence of single site ofrecurrence was observed in the ipsilateral hemithorax in 18patients (44%), contralaterally in 2 (5%), and in the abdomenin 1 patient (2%). Bilateral thoracic relapse was detected inthree patients (7%). Simultaneous recurrence in the ipsilateralhemithorax and abdomen was observed in ten (24%) patientsand in seven (17%) in all three cavities. Unsuspected distantmetastases were detected in 11 patients (26%). Four patternsof uptake were observed in recurrent disease: focal, linear,mixed (focal/linear), and encasing, with a significant differ-ence between the intensity of uptake in malignant lesionscompared to benign post-therapeutic changes. Lesion uptakewas lower in patients previously treated with more aggressivetherapy and higher in intrathoracic lesions of patients withdistant metastases. FDG PET/CT helped in the selection of 12patients (29%) who benefited from additional previouslyunplanned treatment at the time of failure. Multivariateanalysis showed that histologic type remained the onlyindependent predictor of progression-free survival. Survivalafter relapse was independently predicted by the pattern ofFDG uptake and PET nodal status, and overall survival by themaximum standard uptake value.Conclusion FDG PET/CT is an accurate modality todiagnose and to estimate the extent of locoregional anddistant MPM recurrence, and it carries independent prog-nostic value. Once the disease recurs, survival outcomesseem to be independent of histologic type and highlydependent on the intensity of lesion uptake and on thepattern of metabolically active disease in FDG PET/CT.Our observations should be considered limited to patientstreated surgically with or without perioperative therapiesand should not be extrapolated to those unresectable casestreated with chemotherapy alone.

V. H. Gerbaudo (*) :M. MamedeDivision of Nuclear Medicine and Molecular Imaging,Brigham & Women’s Hospital, Harvard Medical School,75 Francis St.,Boston, MA 02115, USAe-mail: vgerbaudo@partners.org

B. Trotman-Dickenson :H. HatabuDivision of Thoracic Radiology, Brigham & Women’s Hospital,Harvard Medical School,Boston, MA 02115, USA

D. J. SugarbakerDivision of Thoracic Surgery, Brigham & Women’s Hospital,Harvard Medical School,Boston, MA 02115, USA

Eur J Nucl Med Mol Imaging (2011) 38:810–821DOI 10.1007/s00259-010-1704-x

Keywords FDG . FDG PET/CT.Mesothelioma . Lungcancer . Response to treatment . Recurrence . Survival

Introduction

Locoregional and distant recurrences are common in malignantpleural mesothelioma (MPM). Data from single institutionstudies show local recurrence rates ranging from 12 to 88%,and distant failure rates from 9 to 100% of cases, depending onthe type of surgical resection used, combined or not withadjuvant chemotherapy and radiotherapy [1–9]. Local recur-rence is usually more common following pleurectomy/decortication (PD), whereas a combination of local anddistant failure is more commonly seen after extrapleuralpneumonectomy (EPP) [9, 10].

Currently, CT and MRI are considered the imagingmodalities of choice for the assessment of treatment failureof MPM. Different morphologic patterns of recurrence havebeen described, such as soft tissue masses along theresection margins, new pulmonary nodules, mediastinallymphadenopathy, ascites, and peritoneal dissemination[11]. However, these imaging findings are not alwaysspecific enough to distinguish recurrent tumor from benignpost-therapeutic changes. Granulation tissue along theresection margins can be irregular and nodular, and beconcerning for recurrence in CT. Therefore, fluorodeox-yglucose (FDG) positron emission tomography (PET)/CThas been proposed as a valuable technique worth exploringin this setting [11, 12].

The role of FDG imaging for the diagnosis and initialstaging of MPM has been studied and reported by severalinvestigators with encouraging results [13–18]. Presently,the application of FDG PET and FDG PET/CT for thedetection of recurrent mesothelioma is increasingly beingexplored [19, 20]. We hypothesized that the metaboliccharacterization in terms of patterns and intensity of FDGuptake of MPM recurrence provides prognostic informationfor the management of these patients. Therefore, followingInstitutional Review Board approval, we studied thediagnostic performance and the prognostic value of FDGPET/CT in suspected MPM relapse in the context ofpatterns and intensity of FDG uptake, histologic type, andtreatment algorithm.

Materials and methods

Patients

Fifty consecutive patients (34 men and 16 women, medianage 62 years, range 37–86 years) with a previous historyof histologically confirmed MPM (37 epithelial and 13

non-epithelial), and with clinical and radiographic suspicionof recurrence, underwent FDG PET/CT imaging as part oftheir restaging workup between October 2003 and December2005. Of the 50 patients, 13 also underwent FDG PET/CTimaging for initial staging before therapy.

Previous therapy (Tx) included: PD (n=23) and EPP(n=27). Thirty (9 during PD; 21 during EPP) patients alsoreceived intraoperative, intracavitary hyperthermic lavage(HCh) for 1 h with 225 mg/m2 of cisplatin at 42°C. Theintravenous chemotherapy protocol varied depending on theclinical trial in which the patient was enrolled. Four patientsreceived neoadjuvant chemotherapy with cisplatin andpemetrexed. Thirty-three received adjuvant chemotherapy(AdCh) 4–6 weeks after surgery, which included:cisplatin + pemetrexed (n=19), cisplatin + gemcitabine(n=7), Navelbine (vinorelbine) + pemetrexed (n=3),Navelbine (vinorelbine) + gemcitabine (n=1), pemetrexedalone (n=2), and mitomycin + methotrexate + mitoxantrone(n=1). Fourteen patients also received external beamradiotherapy, consisting of a median hemithoracic doseof 54 Gy. The treatment algorithms were further classifiedinto three groups: single modality therapy (Tx1: PD orEPP ± HCh) (n=18), bimodality therapy (Tx2: PD or EPP ±HCh + AdCh) (n=18), and trimodality therapy (Tx3: PD orEPP ± HCh + AdCh + Rx) (n=14). After treatment, patientswere followed up at 3- to 5-month intervals.

Confirmation of tumor recurrence was established byhistologic or cytologic analysis, or by an average clinicaland imaging (contrast-enhanced CT, FDG PET/CT, andMRI) follow-up of 5.9 months. Specimens were obtainedfrom needle biopsy, thoracoscopy, pleuroscopy, cytologicexamination of pleural fluid, or mediastinoscopy.

FDG PET/CT imaging

FDG PET/CT scans were performed 11±6 months aftertherapy. After fasting for a period of 6 h, serum glucoselevels were measured in all patients, who were then injectedintravenously with 759±121 MBq (20.6±3 mCi) of FDG.An uptake phase of 92.5±20 min post-injection wasallowed prior to imaging. Patients were encouraged to voidbefore imaging and scanned with a combined FDG PET/CTscanner (Discovery ST, General Electric Medical Systems).Patients were imaged in the supine position, with their armsplaced above their heads when possible, and without anyspecific breath-holding instructions. Unenhanced CT scansfor attenuation correction and anatomic coregistration wereperformed first, from the patient’s head to the proximalthighs, using the following acquisition parameters:140 kVp, 75–120 mA (varying according to the patient’sweight), 0.8 s per CT rotation, a pitch of 1.675:1, areconstructed slice thickness of 3.75 mm, for a totalscanning time of 42.4 s. PET emission scans were acquired

Eur J Nucl Med Mol Imaging (2011) 38:810–821 811

in two-dimensional mode starting at the mid-thighs towardthe head, for six to seven bed positions of 4 min each (47image planes/bed position, 15.4-cm longitudinal field ofview). The CT data were reconstructed in a 512×512matrix using a filtered backprojection algorithm. PET datawere reconstructed using an ordered subset expectationmaximization iterative algorithm (30 subsets, 2 iterations),yielding a volume of 47 slices.

Image analysis

Visual and semiquantitative analyses of the CT, attenuation-corrected and uncorrected PET data, and of the FDG PET/CT fused images were performed with the use of adedicated computer workstation (Xeleris, General ElectricMedical Systems, Haifa, Israel). Two reviewers experiencedin FDG PET and one in CT imaging interpreted the studies,blinded to the clinical findings, to the histopathologicresults, and to previous imaging findings. The results weredocumented by consensus.

A study was considered positive when findings consistentwith recurrent disease were detected in the FDG PET and/orCT images. The FDG PET and fused FDG PET/CT imageswere considered positive for recurrence by the presence of atleast one focal, linear, or encasing area of increased uptakecompared to that of surrounding tissues in regions unrelated tobenign or physiologic processes observed in the CT images.The degree of FDG avidity by each lesion was analyzedsemiquantitatively by obtaining average and maximumstandard uptake values (SUVavg and SUVmax) corrected forbody weight, from a 1.5-cm circular region of interest drawnon the slice in which the lesion had the highest radioactivityconcentration. CT images were considered positive forrecurrence when there was evidence of circumferentialpleural thickening, nodularity or irregularity of thepleural contour, mediastinal pleural involvement, and/orsolitary or multifocal areas of disease infiltrating thechest wall or diaphragm. The criteria used to define abnormallymphadenopathy on CT images varied according to specificmediastinal regions: (a) ≥13 mm short axis diameter for nodesin the subcarinal, precarinal, and tracheobronchial regions and(b) >10 mm for the other nodal stations; (c) the presence ofinternal mammary nodes were regarded as abnormal; (d)diaphragmatic and pericardial nodes measuring ≥5 mm wereconsidered abnormal; and (e) the presence of retrocrural orextrapleural nodes was also considered abnormal [21].Considering that there is not a well-defined and acceptedsize criterion for the imaging characterization of nodaldisease in MPM, and that the accuracy of both CT andPET is known to be far from perfect, we chose to adopt thecriteria introduced by Ikezoe et al. The investigatorsevaluated the use of the above criterion to assess thesensitivity of CT for detecting mediastinal lymph node

metastases in 208 cases of surgically proven non-small celllung cancer (NSCLC). The use of a short-axis nodaldiameter of 13 mm as the upper limit for normal nodes inthe subcarinal, precarinal, and tracheobronchial regions, andof 10 mm for all other nodal stations, resulted in a significantreduction of false-positive findings and higher accuracywhen compared with the criterion of 10 mm diameter for allmediastinal nodes [21]

Statistical analysis

Results were expressed as the mean ± SD. Within-patientcomparisons of metabolic activity as a function of histo-logic type, treatment algorithm, and time to progressionwere performed using parametric and nonparametric testsas appropriate, with p<0.05 used to define statisticalsignificance.

The diagnostic performance of FDG PET/CT to detectdisease recurrence on a per-patient basis was expressed interms of sensitivity, specificity, accuracy, and positive andnegative predictive values using standard formulae.

Overall survival time was defined as the time betweeninitial treatment and death or the last follow-up date.Progression-free survival (PFS) was defined as the timebetween treatment and the earliest clinical evidence ofdisease recurrence. Patients that died without clinical orimaging evidence of relapse were censored on the date ofthe most recent evaluation when recurrence was notevident. Death after documented recurrence was consideredrelated to it. Survival after FDG PET/CT was defined as thetime between the scan and death or last follow-up.

SUV cutoff values used for survival analysis weredefined by receiver-operating characteristic (ROC) analysis,with the area under the ROC curve being determinedwith its corresponding 95% confidence interval (CI).Kaplan-Meier plots were used to estimate cumulativesurvival and a log-rank test to estimate the statisticalsignificance of the difference between groups. Theprognostic significance of all variables under study wasfurther tested in multivariate analysis using Cox proportionalhazards regression analysis with forward exclusion at asignificance level of 0.10.

Results

FDG PET/CT diagnostic performance and sitesof recurrence

Treatment failure was confirmed in 42 patients (84%) [30epithelial and 12 non-epithelial (11 mixed and 1 sarcomatoid)].In 22 patients (52%) the final diagnosis was confirmed byhistology, cytology, and autopsy and in the remaining 20 (48%)

812 Eur J Nucl Med Mol Imaging (2011) 38:810–821

by subsequent imaging and clinical follow-up (5.9 months). Ineight patients, recurrence was excluded by a clinical andimaging median follow-up of 21 months.

Of 42 patients with recurrence, 41 had a positive FDGPET/CT, and the absence of disease was confirmed in 6patients with a negative scan. FDG PET/CTwas true-positivein 41, true-negative in 6, false-positive in 2, and false-negativein 1 case. Overall sensitivity, specificity, accuracy, negativepredictive value, and positive predictive value were 97.6%(95%CI 86–99%), 75% (95%CI 36–99%), 94% (95%CI 87–99%), 86% (95% CI 72–94%), and 95.3% (95% CI 83–99%),respectively.

The false-negative case was an epithelial subtype MPM.Follow-up diagnostic CT at 3.5 months demonstratedprogression evidenced by ascites coupled with peritonealnodular tumor growth. One of the two false-positive studiesshowed moderate increased 18F-FDG avidity (SUVmax

5.51) in the right cardiophrenic angle. Lack of recurrencewas verified at autopsy 3 months later, with radiationpneumonitis confirmed as the cause of death. The otherfalse-positive case showed an intense linear focus of uptake(SUVmax 9.05) extending to the right chest wall, which wasconfirmed by biopsy as a bronchopleural fistula withempyema.

FDG PET/CT evidence of a single site of recurrence wasobserved in 21 patients [ipsilateral hemithorax in 18patients (44%), contralateral hemithorax in 2 (5%), andabdomen in 1 (2%)]. Bilateral thoracic relapse was detectedin three patients (7%), whereas simultaneous recurrenceinvolving the ipsilateral hemithorax and the abdomen wasobserved in ten (24%). In seven patients (17%) the diseasewas detected in all three cavities. FDG PET/CT demon-strated unsuspected distant metastases in 11 patients (26%)(7 epithelial and 4 non-epithelial), with a total of 20 lesions(hematopoietic 11, lymphatic 9).

Twenty patients (74%) recurred after EPP compared to22 (96%) after PD. Ipsilateral recurrence was more oftenseen after PD (59%) than EPP (30%), whereas distantrelapse was more frequent after EPP (35%) than PD(18.2%) (p=0.04). The number of recurrent sites was lowerin patients that had undergone more aggressive treatment.

FDG PET/CT patterns and intensity of uptake

Intrathoracic recurrence

Intrathoracic sites of recurrence observed in FDG PET/CTimages included the chest wall (n=20), mediastinal organs (n=10), intrapleural nodes (N1) (n=1), extrapleural (N2) (n=17)and contralateral (N3) nodes (n=8), the pleura or neopleura(n=5), and the diaphragm or diaphragmatic graft (n=27).

Four distinct patterns of FDG uptake were observed in theconfirmed recurrent lesions: focal (PET-T1) (n=12), linear

(PET-T2) (n=3), mixed (focal/linear) (PET-T3) (n=8), andencasing (PET-T4) (n=18) (Fig. 1). In patients with chestwall relapse, FDG PET/CT showed different patterns ofuptake depending on how advanced the disease was. ThePET-T1 pattern was congruent with solitary foci of tumor,whereas the PET-T2 pattern was consistent with the presenceof multifocal discrete areas of tumor extending into the chest.The PET-T3 pattern was seen in those cases in which the CTimages revealed more advanced local recurrence, diffuselyinvading the intercostal or chest wall muscles.

The pattern of uptake in pleural disease was focal (PET-T1) in areas of thickening or nodularity, or linear (PET-T2)along the lateral, medial, and/or basal aspects of the pleura,with or without diffuse extension to the adjacent lungparenchyma. Uptake in the basal pleura could not bedistinguished from diaphragmatic uptake. Involvement ofthe diaphragm or the reconstructed diaphragm could onlybe unequivocally ascertained when transdiaphragmaticspread was evident.

When the disease involved the mediastinal organs, FDGuptake was diffuse and ill-defined (PET-T3), and it was notpossible to distinguish lesions affecting individual struc-tures based on the FDG PET images alone. In these casesthe CT images complemented the metabolic findings byproviding the necessary anatomic landscape to identifycompromise of each mediastinal structure. In those caseswith more extensive recurrent disease involving the entirehemithorax, the observed pattern of FDG uptake wasencasing (PET-T4), corresponding to areas of circumferentialpleural or tissue thickening seen on CT images.

There was a significant difference between the intensity ofradiotracer uptake in malignant lesions compared to benignpost-surgical changes in relapse-free patients [SUVmax 11.27±4.76, range 2–22.7 (n=42) versus 3.28±1.67, range 2.1–6.96(n=8), respectively; p=0.0001].

The intensity of uptake in intrathoracic recurrent lesions(SUVmax 11.27±4.76) was higher than in lesions detectedat initial staging (SUVmax 9.13±4.90); however, thedifference was not significant. Before therapy, the intensityof uptake was higher in non-epithelial tumors (SUVmax

12.26±4.29) when compared to the epithelial subtype(SUVmax 7.17±4.37), with the difference showing a trendtowards significance (p<0.06). Upon recurrence the differ-ence in intensity of metabolic activity between histologictypes was no longer evident (epithelial SUVmax 11.7±4.9versus non-epithelial SUVmax 10.29±4.68). However,lesion uptake increased from 9.8±5 in patients with singlecavity disease (n=18) to 15.3±4.73 in those that recurred inthree cavities simultaneously (n=7) (p=0.01) (Table 1).FDG avidity was consistently higher in the intrathoraciclesions of patients with distant metastases than in patientswithout [SUVmax 15.4±3.15 (n=11) versus 8.59±4.9 (n=31); p=0.0001].

Eur J Nucl Med Mol Imaging (2011) 38:810–821 813

Of the 42 patients that relapsed, 26 (62%) had FDG PET/CTevidence of nodal recurrence (N1=1, N2=17, and N3=8).Those with epithelial histology had evidence of 1 N1, 11 N2,and 6N3, and the non-epithelial cases had 6 N2 and 2N3. FDGPETevidence of recurrent nodal disease decreased as a functionof more aggressive treatment [after Tx1: 13 (77%) patients;Tx2: 9 (47%); and Tx3: 4 (29%)]. The average N-SUVmax was7.6±4.6 (N1: 2.8, N2: 5.9±1.7, and N3: 11.8±6.0; p=0.001).The average CT size of FDG-positive nodes was 1.58±0.56 cm; range 1–3.5 cm (N1: 2 cm, N2: 1.4±0.29 cm, andN3: 1.8±0.86 cm), with no correlation found between thedegree of uptake and nodal size (r: 0.28; p=0.21).

The metabolic activity of recurrent lesions was lower inpatients that had been treated with more aggressive therapy[after Tx1: SUV: 13.5±4.48 (n=16), Tx2: SUV: 10.75±4.85(n=15), and Tx3: 8.7±3.75 (n=11)]. The difference inrecurrent lesions SUVmax only approached significance whencomparing single to trimodality therapy algorithms (p=0.007).

Abdominal recurrence

FDG PET/CT evidence of abdominal recurrence includedtransdiaphragmatic spread of tumor (n=17), peritoneal nodules(n=9), abdominal wall disease (n=6), and ascites (n=2).

Fig. 1 Axial CT, FDG PET, and FDG PET/CT images illustrating thepatterns of FDG uptake in recurrent MPM. Row a: focal pattern ofintense FDG uptake corresponding to right anterobasal pleuralthickening; row b: linear pattern of moderate FDG uptake along theright lateral chest wall, consistent with pleural thickening on CT; row

c: mixed pattern of FDG uptake in right hemithorax with posteriorchest wall invasion, and row d: encasing pattern of FDG uptake inextensive right circumferential nodular pleural thickening, with masseffect causing leftward displacement of the mediastinum

814 Eur J Nucl Med Mol Imaging (2011) 38:810–821

The SUVmax of transdiaphragmatic spread was 13.66±4.05 (n=17). A commonly observed pattern was a highlyFDG-avid soft tissue thickening of the thoracoabdominalwall extending from the pleural space into the abdomen(Fig. 2).

Systemic metastasis

FDG PET/CT evidence of unsuspected distant metastaseswas observed in 11 patients (26%) with a total of 20 lesions(hematopoietic = 11 and lymphatic = 9). Seven patients hadepithelial and four non-epithelial tumors. Four patients hadbeen treated with pleurectomy and seven with EPP. Allpatients with evidence of metastatic disease in FDG PET/CT had additional findings consistent with nonsystemicrelapse. Two patients had intrathoracic disease; one had

abdominal lesions; two had bilateral thoracic recurrence;three had simultaneous ipsilateral and abdominal disease;and the remaining three had evidence of relapse in all threecavities.

Hematopoietic metastases were detected in the brain (n=1),in the liver (n=3), in the adrenal gland (n=3), and in bone[n=4 (3 lytic; 1 sclerotic)] (Fig. 3) (Table 2).

FDG uptake in lymphatic metastases was observed inretroperitoneal nodes (n=5), in celiac axis nodes (n=2), inone axillary, and one supraclavicular node (Table 2).

The number of systemic sites of relapse was lower inpatients that had received more aggressive treatment.Hematologic metastases were detected in 6 of 16 patients(38%) that recurred after single modality therapy comparedto 1 of 11 (9%) patients that relapsed after trimodalitytreatment. Although not significant, there was a trend for alower incidence of metastatic disease in those patients thatreceived intracavitary hyperthermic lavage as part of theirtrimodality therapy.

Survival

Progression-free survival

The median PFS among the 50 patients was 6.5 months(range 1–47 months). In univariate analysis, epithelialhistology, lack of nodal disease at presentation (pN0), alesion SUVmax <10.7, and PET-T1 and T2 patterns ofuptake at initial staging were significantly associated withprolonged recurrent-free survival (Table 3). Multivariateanalysis showed that histologic type (relative risk: 9.42; p=0.047) was the only independent predictor of PFS in ourpatient population (Table 3) (Fig. 4a).

Survival after FDG PET/CT for restaging

Median survival time after relapse was 6 months (range 1–48 months). In univariate analysis, a negative FDG PET,PET-T1 (focal) and T2 (linear) patterns of lesion uptake,lack of nodal uptake, and a recurrent lesion SUVmax <3

Fig. 2 Pattern of FDG uptake in recurrent transdiaphragmatic disease.Coronal FDG PET and FDG PET/CT images of a 61-year-old womanstatus post right EPP, with fluid filling the pneumonectomy space. Inthe left hemithorax, there is an encasing pattern of FDG uptakecorresponding to circumferential pleural thickening seen on CT.Increased uptake is also present along the peritoneal surfaces,surrounding the spleen, in bilateral paracolic gutters, insinuatingbetween bowel loops, and extending to the deep pelvis

Single site Bilateral Three cavities

Ipsilateral hemithorax Thorax

n=18 (44%) n=3 (7.3%) n=7 (17%)

SUVmax 9.8±5 SUVmax 12.7 ±3 SUVmax 15.3±4.7

Contralateral hemithorax Ipsilateral thorax + abdomen

n=2 (5%) n=10 (24.3%)

SUVmax 11.8±1.3 SUVmax 13.1±2.4

Abdomen

n=1 (2.4%)

SUVmax 13

Table 1 Sites and metabolicactivity of nonsystemic recurrentlesions

n = number of patients with apositive FDG PET/CT

Eur J Nucl Med Mol Imaging (2011) 38:810–821 815

were significantly associated with prolonged survival afterFDG PET imaging (Table 4). Multivariate analysis revealedthat the only independent predictors of survival followingrestaging FDG PET/CT were the intrathoracic PET-Tpattern of uptake (relative risk: 2.34; p=0.01) and thePET-N status (relative risk: 2.26; p=0.05) (Table 4)(Fig. 4b, c).

Overall survival

Among the 50 patients studied, 34 (68%) are deceased, 15(30%) patients are known to be alive, and 1 (2%) was lostto follow-up. The median survival time among 49 patientswas 21 months (range 4–88 months). In univariate analysis,epithelial histology, lack of intrapleural (N1) and extrap-leural (N2) nodal disease, a lesion SUVmax <10.7 at initialstaging, and trimodality therapy were significantly associ-ated with prolonged survival (Table 5). Multivariateanalysis revealed that the lesion SUVmax at initial staging(relative risk: 7.78; p=0.017) was the only independentpredictor of overall survival in our cohort (Table 5)(Fig. 4d).

Impact on management

FDG PET/CT had an impact on the management of 12patients (29%) with confirmed failure, by initiating previ-ously unplanned treatment (chemotherapy = 5, radiation =3, chemotherapy + radiation = 3, and resection +

chemotherapy = 1). Although not significant, these patientshad a longer survival after relapse (median 13 months)compared to those that recurred but did not receiveadditional treatment (n=30; median 6 months) (p=0.26).

Discussion

Our study cohort carried a high clinical suspicion ofrecurrent disease, but it represented the typical clinicallyindeterminate scenarios of daily MPM practice presentingwith inconclusive conventional imaging findings. Ourresults provide encouraging evidence that FDG PET/CT isan accurate imaging technique to detect and stage MPMrecurrence, and that it carries independent prognostic value.

As we have previously described at diagnosis and initialstaging of MPM, four distinct patterns of uptake wereobserved in recurrent lesions: focal, linear, mixed (focal +linear), and encasing, depending on how advanced thedisease was [16, 17]. These patterns and the intensity oflesion uptake correlated well with and confirmed malig-nancy in the CT abnormalities observed after therapy. Insome cases, the imaging findings from CT alone proved tobe unreliable to distinguish recurrent tumor from benignpost-therapeutic changes. In contrast, we found a significantdifference in the intensity of radiotracer uptake betweenmalignant lesions and benign changes in relapse-freepatients. FDG PET/CT was true-positive for recurrence in41 cases and true-negative in 6, yielding only 1 false-

Hematopoietic n SUVmax ± SD (range) Lymphatic n SUVmax ± SD (range)

Brain 1 12.4 Supraclavicular 1 8.7

Liver 3 8.5±4.7 (5.7–15.5) Axillary 1 8.5

Adrenal 3 6.8±1.9 (5.3–8.90) Celiac axis 2 19.3 and 9.9

Bone 4 8.6±3.9 (5.4–14.2) Retroperitoneal 5 9.5±1.8 (7.3–12)

Table 2 FDG PET/CT evidenceof distant systemic failure (11patients)

816 Eur J Nucl Med Mol Imaging (2011) 38:810–821

Fig. 3 Axial FDG PET/CT images of systemic recurrence. a Intensering-like focus of FDG uptake (SUVmax = 12.4) corresponding to a4-cm left posterior parietal metastasis. b Focus of intense FDG uptake(SUVmax = 5.7) in a left hepatic lobe metastatic lesion. In addition,there is intense FDG uptake in the thickened posterobasal aspect of theleft pleura. c Moderate FDG activity (SUVmax = 4.21) in a 1.4-cm

right adrenal nodule, consistent with biopsy-proven adrenal metastasis.There is also intense FDG uptake corresponding to an area of pleuralthickening in the posterior aspect of the left pleura extending to thediaphragm. d Focus of intense FDG uptake (SUVmax = 5.4) in a 1-cmlytic metastatic lesion to the left sacrum

negative and 2 false-positive findings (sensitivity 97.6%,specificity 75%, and accuracy 94%). The false-negativefinding was a non-FDG-avid epithelial MPM confirmed asrecurrent disease by CT 3.5 months later, which showedascites and small peritoneal nodules. Although mostmesotheliomas are highly FDG avid, lack of tracer uptakewas previously reported in the epithelial subtype [13, 15].The first false-positive study showed moderate increasedFDG activity in the right cardiophrenic angle, confirmed asradiation pneumonitis at autopsy 3 months later. IncreasedFDG uptake is not uncommon in inflammatory andinfectious processes, and radiation-induced pneumonitis isa known cause of false-positive findings in FDG PET [22–29]. The other false-positive case showed an intense linearfocus of uptake extending to the right chest wall, confirmedby biopsy as a bronchopleural fistula with empyema.Bronchopleural fistulas result from failure of the bronchialstump to heal due to improper initial closure, inadequateblood supply, residual malignant tumor, or infection at thebronchial stump [30]. Thoracotomy and preoperativechemotherapy and radiation increase the risk of broncho-pleural fistula formation. The recruited white blood cells tothese sites tend to be FDG avid, because they express thenecessary rate-limiting steps for radiotracer uptake (GLUTreceptors and hexokinase) [31, 32]. The diagnostic accura-cy of FDG PET/CT observed in our study comparesfavorably with the results of Tan and colleagues [20], whorecently published the FDG PET/CT findings of a retro-spective study on 25 patients after multimodality treatmentof MPM. The investigators reported a sensitivity of 94%,and a specificity of 100%, of FDG PET/CT to diagnose thepresence of MPM recurrence. In their study, they had onlyone false-negative case, which biopsy confirmed as smallnodules of recurrent MPM in the parietal and visceralpleura.

Aggressive therapies for MPM have extended survivaland, therefore, the potential to develop distant metastases

[33]. A well-documented advantage of FDG PET overanatomic imaging modalities is its greater accuracy for thedetection of systemic metastases [13, 14, 16, 18]. As shownby Erasmus and colleagues, FDG PET/CT improvesextrathoracic staging by detecting lesions missed inconventional imaging or by correctly characterizing equiv-ocal lesions [18]. Erasmus et al. indicated that preoperativeFDG PET/CT detected extrathoracic metastases not identi-fied by conventional imaging in 24% of patients withMPM. In our study, unsuspected distant metastatic recur-rence was detected in 11 patients (26%), with 20 lesions.As we have described at initial staging of MPM, and lateron confirmed by Lee et al., metastatic failure sites weremore FDG avid than primary lesions, and the metabolicactivity in primary lesions of patients with metastases wassignificantly higher than in those without [17, 22, 34].These findings tend to support our previously proposedhypothesis that tumors with high metastatic potential havehigher energy requirements, suggesting that the observedincreased glycolytic activity in primary lesions might be anecessary precondition for the acquisition of metastaticpotential [17, 22]. The number of systemic sites of relapsewas lower in those patients that had received moreaggressive treatment. Metastases occurred in 38% ofpatients that were treated with surgery alone compared toonly 9% who relapsed after trimodality therapy.

The median PFS of the study population was 6.5 months(range 1–47 months). Non-epithelial histology, confirmednodal disease at presentation, a preoperative primary lesionSUVmax >10.7, and a mixed or encasing pattern of uptakewere significantly associated with a shorter recurrent-freeinterval. However, on multivariate analysis the histologic typeremained the only independent predictor of PFS (epithelial:10 months versus non-epithelial: 4 months; p=0.01). Aftertreatment, patients were followed up with 3- to 5-monthsintervals, and the heterogeneity of the surveillance intervalcould have led to bias on the PFS estimates. Therefore, in

Table 3 Predictors of progression-free survival (PFS)

Variable Groups PFS (median = months) PFS rate at 4 months p Log-rank p Cox regression(relative risk)

Histology Epithelial 10 75% 0.01 0.047 (9.42)Non-epithelial 4 39%

Pathology N status Negative 23 72% 0.05 a

Intrapleural (N1) 10 67%Extrapleural (N2/N3) 4 48%

SUVmax at initial staging <10.7 10 89% 0.02 a

≥10.7 4 20%

PET pattern at initial staging Focal/linear Not reached 100% 0.05 a

Mixed/encasing 4 38%

aVariables excluded by the Cox model as they were not found to significantly contribute to the prediction of PFS

Eur J Nucl Med Mol Imaging (2011) 38:810–821 817

addition to median PFS, we also calculated and reported a4-month PFS rate (Table 3). This binary, fixed time endpointallows for more meaningful and uniform comparisons ofmedian PFS across patients and studies [35]. Our multivariateanalysis results complement those from univariate analysisrecently published by Lee and colleagues, who reported ashorter disease-free interval in patients with mixed histology(4 months) compared to those with epithelial subtype tumors(7 months) (p=0.015) [34].

A trend for a longer recurrence-free interval wasobserved in those cases treated with pleurectomy-basedtrimodality therapy combined with intracavitary hyperthermiclavage. These results are in line with those of Richards and

colleagues [36]. In their recent phase I to II study of PDand intraoperative intracavitary hyperthermic cisplatinlavage for MPM, the investigators found a significantdifference in recurrence-free interval for resectablepatients exposed to low doses (median 4 months) versushigh doses (median 9 months) of hyperthermic cisplatin(p<0.0001).

The intensity and extent of metabolically activedisease in primary lesions before therapy were higherin non-epithelial tumors. As lesions recurred, thesemetabolic differences between histologic types were nolonger evident. Furthermore, uptake in recurrent lesionswas significantly higher as the number of cavities

Histologic Type

0 10 20 30 40 50

100

80

60

40

20

0

Time (months)

Pro

gres

sion

-Fre

e S

urvi

val p

roba

bilit

y (%

)a

EpithelialNon-Epithelial

P = 0.01

b PET Pattern of Recurrence

0 10 20 30 40 50

100

80

60

40

20

0

Time (months)

Sur

viva

l pro

babi

lity

afte

r R

esta

ging

PE

T/C

T (

%)

P = 0.001

NegativeFocal/LinearMixed/ Encasing

cRestaging PET-N Status

0 10 20 30 40 50

100

80

60

40

20

0

Time (months)

Sur

viva

l pro

babi

lity

afte

r R

esta

ging

PE

T/C

T (

%)

P = 0.0004

N (-)N (+)

d SUVmax

0 10 20 30 40 50

100

80

60

40

20

0

Time (months)

Ove

rall

Sur

viva

l pro

babi

lity

(%)

P= 0.005

< 10.7>_ 10.7

Fig. 4 Kaplan-Meier plots of the independent predictors of survival.a Log-rank comparison demonstrating significantly longer PFS forpatients with epithelial versus non-epithelial tumors. b Significantlyshorter survival after FDG PET/CT for those patients with mixed andencasing patterns of FDG uptake compared to those with focal/linear

uptake, or with a negative FDG PET/CT scan. c Significantly longersurvival for those patients without FDG PET/CT evidence of nodaldisease [N (-)] versus those with positive nodes [N (+)]. d Influence ofintensity of primary lesion FDG uptake on the overall survival ofmesothelioma patients

818 Eur J Nucl Med Mol Imaging (2011) 38:810–821

affected increased, regardless of the histologic subtype.The SUVmax of ipsilateral intrathoracic lesions was 9.8±5in patients with single cavity disease compared to 15.3±4.73 in those that recurred in the three cavities simulta-neously. A possible explanation for this finding could bethe emergence of more aggressive clones with higherenergy requirements. Interestingly, histology ceased to bea significant predictor of survival following failure, withthe extent of metabolically active tumor (PET pattern) andthe FDG PET nodal status becoming the only independentpredictors of survival after FDG PET evidence of relapse(Table 4) (Fig. 3).

Recurrent lesions were less metabolically active inpatients that had received more aggressive therapy. Thisdifference in recurrent lesion SUVmax approached signifi-cance when comparing single to trimodality therapyalgorithms.

Univariate analysis showed that histology, extrapleuralnode status, lesion SUVmax at initial staging, and trimodalitytherapy were significant predictors of overall survival.Multivariate analysis revealed that a lesion SUVmax >10.7was the only independent predictor of overall survival in ourcohort. Our findings are consistent with those of Ceresoliet al. [37], as well as Flores and colleagues [38], whoreported that a high SUV, mixed histology, and advancedanatomic stage were poor risk factors in MPM. In addition,our findings in the setting of recurrent disease reinforce theresults of Nowak et al., who showed that tumor volume andits glycolytic metabolism may be better predictors of diseaseaggressiveness in mesothelioma [39].

It might be argued that recurrent MPM has a very poorprognosis due to the limited therapeutic options available.Upon failure there is no standard treatment approach, andmanagement is decided and implemented on a patient-by-

Table 5 Predictors of overall survival (OS)

Variable Groups OS (median = months) p Log-rank p Cox regression (relative risk)

Histology Epithelial 29 0.008 a

Non-epithelial 9

Pathology N status Negative 54 0.047 a

Intrapleural (N1) 16Extrapleural (N2/N3) 14

Tumor SUVmax at initial staging <10.7 41 0.005 0.017 (7.78)≥10.7 7

Initial treatment EPP a

Single 9

Bimodality 17

Trimodality 29 0.01

Pleurectomy (P/D) 29 0.01 a

Single 9

Bimodality 53

Trimodality Not reached 0.01

a Variables excluded by the Cox model as they were not found to significantly contribute to the prediction of OS

Table 4 Predictors of survival after restaging PET/CT

Variable Groups Survival after restaging PET(median = months)

p Log-rank p Cox regression(relative risk)

PET pattern of recurrence Negative 39 0.001 0.01 (2.34)Focal/linear 21

Mixed/encasing 5

Restaging PET-N status Negative 32 0.01 0.05 (2.26)Intrapleural (N1) 7Extrapleural (N2/N3) 5

Recurrence SUVmax <3 Not reached 0.01 a

≥3 6

a Variables excluded by the Cox model as they were not found to significantly contribute to the prediction of survival

Eur J Nucl Med Mol Imaging (2011) 38:810–821 819

patient basis. At our institution some patients with localizedchest wall recurrence or lymph node disease are treatedsurgically, while those with more extensive relapse can betreated either with surgery or with chemotherapy orradiation, depending on the lesion and on the patient.Recent studies have shown that these patients can be treatedwith agents not previously employed in the initial treatmentalgorithm [40]. Patients who are chemotherapy naïve maybe candidates for first-line cisplatin and pemetrexed orcisplatin and raltitrexed [41, 42]. Considering that CT aloneoften proves to be unreliable to distinguish recurrent MPMfrom benign post-therapeutic changes, and that PET alonecan be inaccurate to assess N status [18], the good accuracyand early characterization of locoregional and distant failureby hybrid imaging in our cohort helped overcome thelimitations of either modality alone. To this end, our resultsshowed that FDG PET/CT helped in the selection of 12patients (29%) who benefited from additional previouslyunplanned treatment at the time of failure. In these patientsthere was a trend for survival benefit after relapse (median13 months) compared to those that recurred but did notreceive additional unnecessary treatment (n=29; median6 months). Included amongst the latter were 11 patients inwhom FDG PET/CT detected unsuspected metastaticdisease. These encouraging preliminary results merit furtherprospective validation in larger series.

The results of this study demonstrate that FDG PET/CT is an accurate modality to diagnose, and to estimatethe extent of locoregional and distant MPM recurrence,and that it carries independent prognostic value. Oncethe disease recurs, survival outcomes seem to beindependent of histologic type, and highly dependenton the intensity of lesion uptake and on the pattern ofmetabolically active disease in FDG PET/CT. Ourobservations should be considered limited to patientstreated surgically with or without perioperative therapiesand should not be extrapolated to those unresectablecases treated with chemotherapy alone.

Acknowledgement Supported by a grant from the InternationalMesothelioma Program, Brigham & Women’s Hospital, Boston, MA.

Conflicts of interest None.

References

1. Rusch VW, Piantadosi S, Holmes EC. The role of extrapleuralpneumonectomy in malignant pleural mesothelioma: a LungCancer Study Group trial. J Thorac Cardiovasc Surg1991;102:1–9.

2. Pass HI, Kranda K, Temeck BK, Feuerstein I, Steinberg SM.Surgically debulked malignant pleural mesothelioma: results andprognostic factors. Ann Surg Oncol 1997;4:215–22.

3. Hilaris BS, Nori D, Kwong E, Kutcher GJ, Martini N.Pleurectomy and intraoperative brachytherapy and postoperativeradiation in the treatment of malignant pleural mesothelioma. Int JRadiat Oncol Biol Phys 1984;10:325–31.

4. Rusch VW, Rosenzweig K, Venkatraman E, Leon L, Raben A,Harrison L, et al. A phase II trial of surgical resection andadjuvant high-dose hemithoracic radiation for malignant pleuralmesothelioma. J Thorac Cardiovasc Surg 2001;122:788–95.

5. Holsti LR, Pyrhönen S, Kajanti M, Mäntylä M, Mattson K,Maasilta P, et al. Altered fractionation of hemithorax irradiationfor pleural mesothelioma and failure patterns after treatment. ActaOncol 1997;36:397–405.

6. Pass HI, Temeck BK, Kranda K, Thomas G, Russo A, Smith P, etal. Phase III randomized trial of surgery with or withoutintraoperative photodynamic therapy and postoperative immuno-chemotherapy for malignant pleural mesothelioma. Ann SurgOncol 1997;4:628–33.

7. Rusch V, Saltz L, Venkatraman E, Ginsberg R, McCormack P,Burt M, et al. A phase II trial of pleurectomy/decorticationfollowed by intrapleural and systemic chemotherapy for malignantpleural mesothelioma. J Clin Oncol 1994;12:1156–63.

8. Rice TW, Adelstein DJ, Kirby TJ, Saltarelli MG, Murthy SR, VanKirk MA, et al. Aggressive multimodality therapy for malignantpleural mesothelioma. Ann Thorac Surg 1994;58:24–9.

9. Baldini EH, Recht A, Strauss GM, DeCamp MM, Swanson SJ,Liptay MJ, et al. Patterns of failure after trimodality therapy formalignant pleural mesothelioma. Ann Thorac Surg 1997;63(2):334–8.

10. Jänne PA, Baldini EH. Patterns of failure following surgicalresection for malignant pleural mesothelioma. Thorac Surg Clin2004;14(4):567–73.

11. Gill RR, Gerbaudo VH, Sugarbaker DJ, Hatabu H. Current trendsin radiologic management of malignant pleural mesothelioma.Semin Thorac Cardiovasc Surg 2009;21(2):111–20.

12. Yamamuro M, Gerbaudo VH, Gill RR, Jacobson FL, SugarbakerDJ, Hatabu H. Morphologic and functional imaging of malignantpleural mesothelioma. Eur J Radiol 2007;64(3):356–66.

13. Bénard F, Sterman D, Smith RJ, Kaiser LR, Albelda SM, Alavi A.Metabolic imaging of malignant pleural mesothelioma withfluorodeoxyglucose positron emission tomography. Chest1998;114:713–22.

14. Schneider DB, Clary-Macy C, Challa S, Sasse KC, Merrick SH,Hawkins R, et al. Positron emission tomography with f18-fluorodeoxyglucose in the staging and preoperative evaluation ofmalignant pleural mesothelioma. J Thorac Cardiovasc Surg2000;120:128–33.

15. Carretta A, Landoni C, Melloni G, Ceresoli GL, Compierchio A,Fazio F, et al. 18-FDG positron emission tomography in theevaluation of malignant pleural diseases – a pilot study. Eur JCardiothorac Surg 2000;17:377–83.

16. Gerbaudo VH, Sugarbaker DJ, Britz-Cunningham S, Di Carli MF,Mauceri C, Treves ST. Assessment of malignant pleural mesotheliomawith (18)F-FDG dual-head gamma-camera coincidence imaging:comparison with histopathology. J Nucl Med 2002;43:1144–9.

17. Gerbaudo VH. 18F-FDG imaging of malignant pleural mesothelioma:scientiam impendere vero. Nucl Med Commun 2003;24(6):609–14.

18. Erasmus JJ, Truong MT, Smythe WR, Munden RF, Marom EM,Rice DC, et al. Integrated computed tomography-positron emissiontomography in patients with potentially resectable malignant pleuralmesothelioma: staging implications. J Thorac Cardiovasc Surg2005;129(6):1364–70.

19. Fiore D, Baggio V, Sotti G, Muzzio PC. Imaging before and aftermultimodal treatment for malignant pleural mesothelioma. RadiolMed 2006;111(3):355–64.

20. Tan C, Barrington S, Rankin S, Landau D, Pilling J, Spicer J, et al.Role of integrated 18-fluorodeoxyglucose position emission

820 Eur J Nucl Med Mol Imaging (2011) 38:810–821

tomography-computed tomography in patients surveillance aftermultimodality therapy of malignant pleural mesothelioma. JThorac Oncol 2010;5(3):385–8.

21. Ikezoe J, Kadowaki K, Morimoto S, Takashima S, Kozuka T,Nakahara K, et al. Mediastinal lymph node metastases fromnonsmall cell bronchogenic carcinoma: reevaluation with CT. JComput Assist Tomogr 1990;14(3):340–4.

22. Gerbaudo VH, Britz-Cunningham S, Sugarbaker DJ, Treves ST.Metabolic significance of the pattern, intensity and kinetics of18F-FDG uptake in malignant pleural mesothelioma. Thorax2003;58(12):1077–82.

23. Jones HA, Clark RJ, Rhodes CG, Schofield JB, Krausz T, HaslettC. Positron emission tomography of 18FDG uptake in localizedpulmonary inflammation. Acta Radiol Suppl 1991;376:148.

24. Strauss LG. Fluorine-18 deoxyglucose and false-positive results: amajor problem in the diagnostics of oncological patients. Eur JNucl Med 1996;23:1409–15.

25. Zhuang H, Pourdehnad M, Yamamoto AJ, Rossman MD, Alavi A.Intense F-18 fluorodeoxyglucose uptake caused by mycobacteriumavium intracellulare infection. Clin Nucl Med 2001;26(5):458.

26. Alavi A, Gupta N, Alberini JL, Hickeson M, Adam LE, BhargavaP, et al. Positron emission tomography imaging in nonmalignantthoracic disorders. Semin Nucl Med 2002;32(4):293–321.

27. Zhuang H, Duarte PS, Rebenstock A, Feng Q, Alavi A.Pulmonary clostridium perfringens infection detected by FDGpositron emission tomography. Clin Nucl Med 2003;28(6):517–8.

28. Inoue T, Kim EE, Komaki R, Wong FC, Bassa P, Wong WH, et al.Detecting recurrent or residual lung cancer with FDG-PET. J NuclMed 1995;36(5):788–93.

29. Nestle U, Hellwig D, Fleckenstein J, Walter K, Ukena D, Rübe C,et al. Comparison of early pulmonary changes in 18FDG-PET andCT after combined radiochemotherapy for advanced non-small-cell lung cancer: a study in 15 patients. Front Radiat Ther Oncol2002;37:26–33.

30. Sonobe M, Nawkagawa M, Ichinose M, Ikegami N, Nagasawa M,Shindo T. Analysis of risk factors in bronchopleural fistula afterpulmonary resection for primary lung cancer. Eur J CardiothoracSurg 2000;18:519–23.

31. Jones HA, Clark RJ, Rhodes CG, Schofield JB, Krausz T, HaslettC. In vivo measurement of neutrophil activity in experimentallung inflammation. Am J Respir Crit Care Med 1994;149(6):1635–9.

32. Jones HA, Cadwallader KA, White JF, Uddin M, Peters AM,Chilvers ER. Dissociation between respiratory burst activity and

deoxyglucose uptake in human neutrophil granulocytes: implica-tions for interpretation of (18)F-FDG PET images. J Nucl Med2002;43(5):652–7.

33. Zellos L, Christiani DC. Epidemiology, biologic behavior, andnatural history of mesothelioma. Thorac Surg Clin 2004;14(4):469–77.

34. Lee ST, Ghanem M, Herbertson RA, Berlangieri SU, Byrne AJ,Tabone K, et al. Prognostic value of (18)F-FDG PET/CT inpatients with malignant pleural mesothelioma. Mol Imaging Biol2009;11(6):473–9.

35. Panageas KS, Ben-Porat L, Dickler MN, Chapman PB, Schrag D.When you look matters: the effect of assessment schedule onprogression-free survival. J Natl Cancer Inst 2007;99(6):428–32.

36. Richards WG, Zellos L, Bueno R, Jaklitsch MT, Jänne PA,Chirieac LR, et al. Phase I to II study of pleurectomy/decorticationand intraoperative intracavitary hyperthermic cisplatin lavage formesothelioma. J Clin Oncol 2006;24(10):1561–7.

37. Ceresoli GL, Chiti A, Zucali PA, Cappuzzo F, De Vincenzo F,Cavina R, et al. Assessment of tumor response in malignantpleural mesothelioma. Cancer Treat Rev 2007;33(6):533–41.

38. Flores RM, Akhurst T, Gonen M, Zakowski M, Dycoco J, LarsonSM, et al. Positron emission tomography predicts survival inmalignant pleural mesothelioma. J Thorac Cardiovasc Surg2006;132(4):763–8.

39. Nowak AK, Francis RJ, Phillips MJ, Millward MJ, van der SchaafAA, Boucek J, et al. A novel prognostic model for malignantmesothelioma incorporating quantitative FDG-PET imaging withclinical parameters. Clin Cancer Res 2010;16(8):2409–17.

40. Jassem J, Ramlau R, Santoro A, Schuette W, Chemaissani A,Hong S, et al. Phase III trial of pemetrexed plus best supportivecare compared with best supportive care in previously treatedpatients with advanced malignant pleural mesothelioma. J ClinOncol 2008;26(10):1698–704.

41. Vogelzang NJ, Rusthoven JJ, Symanowski J, Denham C, KaukelE, Ruffie P, et al. Phase III study of pemetrexed in combinationwith cisplatin versus cisplatin alone in patients with malignantpleural mesothelioma. J Clin Oncol 2003;21(14):2636–44.

42. van Meerbeeck JP, Gaafar R, Manegold C, Van Klaveren RJ, VanMarck EA, Vincent M, et al. Randomized phase III study ofcisplatin with or without raltitrexed in patients with malignantpleural mesothelioma: an intergroup study of the EuropeanOrganisation for Research and Treatment of Cancer Lung CancerGroup and the National Cancer Institute of Canada. J Clin Oncol2005;23(28):6881–9.

Eur J Nucl Med Mol Imaging (2011) 38:810–821 821