PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of jcoHomeThis ArticleSearchSubmitASCO JCO Homepage
 
J Clin Oncol. May 1, 2012; 30(13): 1541–1549.
Published online Mar 12, 2012. doi:  10.1200/JCO.2011.39.6671
PMCID: PMC3383122
Serum Mesothelin for Diagnosing Malignant Pleural Mesothelioma: An Individual Patient Data Meta-Analysis
Kevin Hollevoet, Johannes B. Reitsma, Jenette Creaney, Bogdan D. Grigoriu, Bruce W. Robinson, Arnaud Scherpereel, Alfonso Cristaudo, Harvey I. Pass, Kristiaan Nackaerts, José A. Rodríguez Portal, Joachim Schneider, Thomas Muley, Francesca Di Serio, Paul Baas, Marco Tomasetti, Alex J. Rai, and Jan P. van Meerbeeck
Author affiliations appear at the end of this article.
Corresponding author: Kevin Hollevoet, PhD, Laboratory of Molecular Biology, Center for Cancer Research, National Cancer Institute, National Institutes of Health, 37 Convent Dr, Room 5110, Bethesda, MD 20892-4264; e-mail: kevin.hollevoet/at/nih.gov.
Received September 26, 2011; Accepted December 13, 2011.
Purpose
Mesothelin is currently considered the best available serum biomarker of malignant pleural mesothelioma. To examine the diagnostic accuracy and use of serum mesothelin in early diagnosis, we performed an individual patient data (IPD) meta-analysis.
Methods
The literature search identified 16 diagnostic studies of serum mesothelin, measured with the Mesomark enzyme-linked immunosorbent assay. IPD of 4,491 individuals were collected, including several control groups and 1,026 patients with malignant pleural mesothelioma. Mesothelin levels were standardized for between-study differences and age, after which the diagnostic accuracy and the factors affecting it were examined with receiver operating characteristic (ROC) regression analysis.
Results
At a common diagnostic threshold of 2.00 nmol/L, the sensitivities and specificities of mesothelin in the different studies ranged widely from 19% to 68% and 88% to 100%, respectively. This heterogeneity can be explained by differences in study population, because type of control group, mesothelioma stage, and histologic subtype significantly affected the diagnostic accuracy. The use of mesothelin in early diagnosis was evaluated by differentiating 217 patients with stage I or II epithelioid and biphasic mesothelioma from 1,612 symptomatic or high-risk controls. The resulting area under the ROC curve was 0.77 (95% CI, 0.73 to 0.81). At 95% specificity, mesothelin displayed a sensitivity of 32% (95% CI, 26% to 40%).
Conclusion
In patients suspected of having mesothelioma, a positive blood test for mesothelin at a high-specificity threshold is a strong incentive to urge further diagnostic steps. However, the poor sensitivity of mesothelin clearly limits its added value to early diagnosis and emphasizes the need for further biomarker research.
Malignant mesothelioma is an asbestos-related malignancy, predominantly arising from the surface serosal cells of the pleura and, to a lesser extent, the peritoneum, pericardium, and tunica vaginalis. The three main histologic subtypes of mesothelioma are epithelioid (60%), sarcomatoid (10%), and biphasic (30%), which combines epithelioid and sarcomatoid features.1 Reported incidences of mesothelioma vary worldwide and are approximately nine per million inhabitants in the United States, 40 per million inhabitants in Australia, and 20 per million inhabitants in Europe, with large between-country differences.2 In most developing countries, epidemiologic data are either unavailable or under-reported.3 In developed countries, peak incidences are expected to occur within the next decade or have been reached already, for example in the United States.2 Mesothelioma, nevertheless, will remain a global health issue for future generations because of the continued use of asbestos in some developing countries, the environmental asbestos exposure, and the long latency period (typically > 30 years) of this malignancy.
Mesothelioma primarily occurs in the older population, and patients currently face a poor prognosis. Current therapeutic options are limited, and mesothelioma is still considered fatal.4 The natural history results in a median survival of 7 to 9 months.2 When treated with standard of care chemotherapy, cisplatin and an antifolate (pemetrexed or raltitrexed), median survival is approximately 1 year.5,6 Highly selected patients with early-stage epithelioid disease, treated with extrapleural pneumonectomy, alone or in combination with chemotherapy and/or radiation therapy, have a median survival of up to 2 years.7
Patients with malignant pleural mesothelioma typically present with symptoms of an underlying pleural effusion, including dyspnea and chest pain.1 The initial diagnostic procedures involve a chest x-ray or computed tomography scan and pleural fluid cytology. The latter may be reliable in experienced hands,8 but a definitive diagnosis typically requires the histologic analysis of pleural biopsies, obtained during thoracoscopy.1,2 Because of the nonspecific presenting symptoms and insidious development of the tumor, mesothelioma is often diagnosed at an advanced stage.1 Because early diagnosis and subsequent intervention are thought to improve disease outcome, there is a critical need for reliable and noninvasive tools that shorten this diagnostic delay.
Because of their noninvasive feature and relative inexpensiveness, serum tumor biomarkers are an attractive adjunct to this purpose. Serum mesothelin, previously referred to as soluble mesothelin-related protein, is currently the most studied and is considered the best available blood protein biomarker of mesothelioma. The mesothelin gene (MSLN) encodes a 69-kDa precursor protein, which is cleaved into a soluble 31-kDa fraction, megakaryocyte potentiating factor, and a membrane-bound 40-kDa glycoprotein, mesothelin.9 The latter is a differentiation antigen that is normally present on the mesothelial cells lining the pleura, peritoneum, and pericardium but is highly expressed in mesothelioma (limited to the epithelioid tumor cells) and some other malignancies, including ovarian and pancreatic cancer.9 Mesothelin has three presumed isoforms that can enter the circulation, either by shedding of the membrane-bound portion (variants 1 and 2) or by a frameshift mutation (variant 3; Appendix Fig A1, online only).10,11 Serum mesothelin refers to all isoforms that are present in the circulation, although variant 1 is predominantly expressed and released from the membrane.11 In 2003, Robinson et al12 were the first to report serum mesothelin as a biomarker of mesothelioma, using an enzyme-linked immunosorbent assay (ELISA) that detects both variants 1 and 3. This assay became later commercialized as Mesomark (Fujirebio Diagnostics, Malvern, PA) and was approved in 2007 by the US Food and Drug Administration to aid in the monitoring of patients with epithelioid and biphasic mesothelioma.13
Serum mesothelin could be an added value to the current diagnostic process if it proves to shorten the diagnostic delay of mesothelioma and lead to an earlier diagnosis. However, despite numerous published studies, the diagnostic use of mesothelin remains under debate. To examine the diagnostic accuracy of mesothelin in the available studies and elucidate whether this biomarker can be an adjunct to an earlier diagnosis of mesothelioma, we performed an individual patient data (IPD) meta-analysis.
Search Strategy and Study Selection
MEDLINE (PubMed database) and EMBASE were searched for studies between 2003, when the first study was published,12 and July 2010 with the following keywords: “mesothelioma” and “mesothelin.” In addition, the references of all publications were manually searched. Meeting abstracts were excluded because of their limited data. Only studies published in peer-reviewed journals and written in English were evaluated for eligibility. To avoid heterogeneity caused by different assay platforms, only studies that measured mesothelin in serum with the commercial Mesomark ELISA kit (Fujirebio Diagnostics) were included. To be eligible, studies also had to include patients with malignant pleural mesothelioma (patient cases) and one or more control groups. Investigators of all eligible studies were invited to join the Mesothelin Collaboration and share the IPD of their patient cases and participants in one of the following five control groups: healthy individuals without asbestos exposure; individuals with reported asbestos exposure and no obvious asbestos-related lesions; individuals with a benign asbestos-related disease; individuals with a benign nonasbestos-related respiratory disease; and individuals with lung cancer. IPD included the serum mesothelin levels (measured in nanomoles per liter), age, and sex of each study participant. In controls, type of control group and, in cases, tumor stage (I or II v III or IV) and histologic subtype (epithelioid, sarcomatoid, or biphasic) were also collected.
Study Quality Assessment
All eligible studies were assessed for methodologic quality using an adapted version of the Quality Assessment of Diagnostic Accuracy Studies tool.14 Each report was evaluated using the following four questions, answered with yes or no: (1) Were the controls representative and well defined? (2) Were the patient cases representative and well defined? (3) Were mesothelin levels measured, blinded to the sample data? (4) Were mesothelin levels measured in the same laboratory? Question 1 was answered positive if the absence of asbestos exposure in healthy individuals, the absence of asbestos-related disease in healthy asbestos-exposed individuals, and the presence of benign asbestos-related diseases were adequately evaluated. Question 2 was scored as positive if patients with mesothelioma were diagnosed according to the reference standards (histopathology or cytology) and were enrolled before any anticancer treatment and if tumor stage and histologic subtype were reported. Questions 1 and 2 also evaluated whether the inclusion of participants was random or consecutive and thus free of selection bias. Questions 3 and 4 were answered positive if measurement bias and between-laboratory variance were avoided.15 If a question could not be answered with the data available in the study, the corresponding author was contacted.
Statistical Analyses
To evaluate whether the distribution of mesothelin levels differed between studies, both in controls and patient cases, the Kruskal-Wallis test was applied. The correlation between age and mesothelin levels of controls was evaluated with the Spearman rank test. To document the between-study heterogeneity in diagnostic accuracy of mesothelin, sensitivity and specificity were determined in each study at a threshold of 2.00 nmol/L, which was arbitrarily chosen from the previously reported diagnostic thresholds.16 The resulting pairs of sensitivity and specificity were meta-analyzed using the bivariate model with a random effects approach, to obtain summary estimates and 95% prediction intervals.17 These predictions intervals show how the sensitivity and specificity of mesothelin are expected to vary in a new study that is comparable in design to the studies included in the meta-analysis. Such a prediction interval is centered at the summary estimate of the sensitivity or specificity, and its width accounts for the uncertainty of the summary estimate, the estimate of between-study variance in true sensitivities or specificities, and the uncertainty in the between-study standard variance estimate itself. The width of these 95% prediction intervals consequently aids to interpret the amount of between-study heterogeneity in the sensitivity and specificity of mesothelin.18 All further analyses were based on IPD and used mesothelin as a continuous variable. Receiver operating characteristic (ROC) regression analysis was performed to examine the diagnostic accuracy of mesothelin and the factors affecting it.19 Before ROC regression, mesothelin levels were standardized for differences in age between patient cases and controls within studies and for differences in the mesothelin levels in controls between studies.19 The resulting regression coefficients were used to fit the ROC curves, of which the area under the curves (AUC), sensitivity, specificity, and likelihood ratios (LRs) could be derived.20,21 Bootstrap resampling was performed (1,000 resamples) to obtain the 95% CIs. The hierarchical nature of the data was preserved by first comparing mesothelin levels between controls and patient cases in each study. Controls of each study were consequently only included in the model if the appropriate patient cases were present, and vice versa. Each group of patient cases or controls had to contain at least 10 individuals. All hypothesis tests were performed two-sided at the 5% significance level. Statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC).
Study Selection
The electronic search identified 189 studies in EMBASE and 119 studies in MEDLINE, resulting in a total of 187 unique studies (Appendix Fig A2, online only). After evaluating the inclusion criteria in title and abstract, 167 studies were excluded, and 20 studies were obtained for full-text evaluation. One study was excluded because of duplicate data,15 and three studies were excluded for using an ELISA other than Mesomark to measure serum mesothelin levels.12,22,23 One study reported mesothelin levels in plasma,24 but the corresponding author provided the matching serum levels. As such, a total of 16 studies were included in the meta-analysis16,2438; for all of these studies, the corresponding authors agreed to share IPD. One research group published four studies,25,27,29,33 and another research group published three studies.26,28,36 After consulting the investigators, their IPD were pooled into one data set per research group, thus excluding duplicate data. This resulted in a total of 11 IPD sets (Table 1).
Table 1.
Table 1.
Studies and Participants Included in the Meta-Analysis
Study Methodology
Question 1 on representative and well-defined controls was negatively scored in two studies,32,34 because the absence of previous asbestos exposure in healthy controls was not ascertained (Appendix Table A1, online only). In the other studies, this was done with a questionnaire. In all asbestos-exposed controls, the presence of a benign asbestos-related condition was radiologically evaluated. Question 2 on representative and well-defined patient cases was negatively scored in six reports, because histologic subtype and, more frequently, tumor stage were not available (Appendix Table A1).25,27,33,24,37,38 One report included a number of patients with mesothelioma with recurrent disease.35 These patients were excluded from the IPD. All 16 studies reported the use of histopathology to diagnose mesothelioma,16,2438 although six of them also used cytology.25,27,29,31,33,37 In two of these studies, cytology was only applied in a small number of patients (five [5%] of 111 patients31 and one [3%] of 36 patients37). In the four other studies,25,27,29,33 published by the same research group, 89 (36%) of 249 patients had a cytology-based diagnosis. Staging was done in accordance with the TNM scoring system of the International Mesothelioma Interest Group39 or the earlier classifications of the Union for International Cancer Control.40 All participants were included in a random or consecutive manner. Questions 3 and 4 on the blinded mesothelin analysis of the samples and the avoidance of interlaboratory variance, respectively, were positively scored in all reports (Appendix Table A1).
Study Population
The IPD of a total of 4,491 participants were obtained; 1,026 patients had malignant pleural mesothelioma and 3,465 participants were controls, including 909 healthy individuals, assumed without asbestos exposure, 775 healthy individuals with reported asbestos exposure, 736 patients with a benign asbestos-related disease (pleural plaques, diffuse pleural thickening, or asbestosis), 267 patients with a benign nonasbestos-related respiratory disease (asthma, chronic obstructive pulmonary disease, or pleural effusion), and 778 patients with lung cancer (non–small-cell and small-cell histology). The size and distribution of these control groups substantially differed across the studies. Healthy individuals, with or without asbestos exposure, were significantly younger than the other groups (P < .001; Table 1). Seventy percent of all participants were men, 15% were women, and data were missing for 15% of participants. Mesothelin levels were available from each participant, whereas data on age, mesothelioma tumor stage, and histologic subtype were missing in 18%, 44%, and 26% of participants, respectively. The majority of the patient cases had epithelioid and advanced stage III or IV mesothelioma (Table 2). Of the patients who lacked data on tumor stage, histologic subtype, or both, the diagnosis was cytology based in none of 201, four (22%) of 18, and 91 (37%) of 249 patients, respectively.
Table 2.
Table 2.
Characteristics of the Patients With Malignant Pleural Mesothelioma (n = 1,026)
Between-Study Heterogeneity
Mesothelin levels differed significantly both within the controls (P < .001) and patient cases (P < .001) of the different studies. In controls, median mesothelin levels varied between 0.34 nmol/L (interquartile range [IQR], 0.23 to 0.53 nmol/L)38 and 0.95 nmol/L (IQR, 0.73 to 1.26 nmol/L),16 whereas in patient cases, median levels ranged between 0.80 nmol/L (IQR, 0.47 to 1.66 nmol/L)37 and 3.41 nmol/L (IQR, 1.62 to 11.73 nmol/L).24 The reported diagnostic thresholds of mesothelin to differentiate controls from patient cases varied widely between 0.93 nmol/L26 and 2.50 nmol/L.27 When applying a common threshold of 2.00 nmol/L, the resulting sensitivities in the different studies ranged from 19%37 to 68%24 (summary estimate, 47%), whereas the specificities varied from 88%25,27,29,33 to 100%38 (summary estimate, 96%; Fig 1). Similarly, the 95% prediction intervals of the sensitivity and specificity of mesothelin ranged widely from 26% to 70% and from 85% to 99%, respectively. Altogether, these findings reflected substantial between-study heterogeneity in the diagnostic accuracy of mesothelin.
Fig 1.
Fig 1.
Forest plots of the (A) sensitivity and (B) specificity of serum mesothelin at a threshold of 2.00 nmol/L. (A) n, patients with mesothelioma with elevated mesothelin levels; N, all patients with mesothelioma. (B) n, controls with normal mesothelin levels; (more ...)
ROC Regression Analysis
Besides differing among studies, mesothelin levels of controls were also significantly correlated with age (r = 0.24; P < .001). Therefore, before the ROC regression analysis, mesothelin levels were standardized for these factors. Because of missing data on age, two complete study populations (n = 536),37,38 a group of healthy controls (n = 409),34 and a group of healthy asbestos-exposed individuals (n = 113)26 were omitted from the ROC regression analysis. One group of healthy asbestos-exposed individuals was excluded because of its limited size (n = 4; Table 1).34 In total, 2,578 (74%) of 3465 controls and 851 (83%) of 1,026 patients with mesothelioma were available for the ROC regression analysis. Results indicated that the type of control group had a significant effect on the diagnostic accuracy of mesothelin levels (Fig 2). The highest AUCs were observed for differentiating patient cases from the two groups of healthy controls, either with or without asbestos exposure (AUC, 0.84; 95% CI, 0.81 to 0.87). Overall, the differences between the AUCs in the four control groups with no malignant disease were relatively modest. The lowest AUC was obtained when differentiating patient cases from patients with lung cancer (AUC, 0.76; 95% CI, 0.73 to 0.79; Fig 2). In addition, tumor stage (I or II v III or IV) and histologic subtype (epithelioid, sarcomatoid, or biphasic) significantly affected the diagnostic accuracy of mesothelin (Table 3). The highest AUC was observed for differentiating patients with epithelioid stage III or IV mesothelioma from controls (AUC, 0.84; 95% CI, 0.82 to 0.86). The lowest AUC was obtained for sarcomatoid stage I or II mesothelioma (AUC, 0.56; 95% CI, 0.51 to 0.60).
Fig 2.
Fig 2.
Receiver operating characteristic curves of standardized serum mesothelin levels for differentiating each control group from patients with malignant pleural mesothelioma. AUC, area under the curve.
Table 3.
Table 3.
AUC of Standardized Serum Mesothelin Levels for Differentiating All Control Groups From Patients With Malignant Pleural Mesothelioma,Stratified According to Histology and Tumor Stage
Mesothelin in Early Diagnosis
The use of mesothelin in early diagnosis was examined by constructing a clinically relevant ROC regression model, again with respect to the hierarchical structure of the data. This model included 217 patients with stage I or II histologically proven epithelioid (n = 185) or biphasic (n = 32) mesothelioma and 1,612 symptomatic or high-risk controls, including 318 healthy asbestos-exposed individuals, 480 patients with a benign asbestos-related disease, 83 patients with a benign nonasbestos-related respiratory disease, and 731 patients with lung cancer. These 1,829 individuals were retrieved from nine studies (Appendix Table A2, online only).16,26,28,3032,3436 Healthy individuals without asbestos exposure were not included, because the use of mesothelin in these individuals is not clinically relevant. When differentiating patient cases from controls, mesothelin levels displayed an AUC of 0.77 (95% CI, 0.73 to 0.81), representing the overall diagnostic performance. In addition, mesothelin was evaluated in the following two specific settings: as an adjunct to rule in (through a positive blood test) or to rule out (through a negative blood test) mesothelioma diagnosis. For a positive test to do so, a high-specificity threshold is typically required. Because mesothelin levels were standardized for between-study differences and age, no thresholds in nanomoles per liter could be derived. Therefore, we opted for a specificity of 95% (ie, a false-positive rate of one out of 20), which resulted in a sensitivity of 32% (95% CI, 26% to 40%) and a positive LR of 6.40. For a negative test result to aid in excluding diagnosis, a high-sensitivity threshold is generally required. At a selected sensitivity of 95%, specificity was 22% (95% CI, 15% to 29%), yielding a negative LR of 0.23. Using the associated LRs, Table 4 illustrates how a mesothelin blood test result shifts the post-test probability of mesothelioma in two hypothetical patients with a pretest probability of 25% and 50%, respectively.
Table 4.
Table 4.
Effect of a Serum Mesothelin Test Result on the Post-Test Probability of Malignant Pleural Mesothelioma in Two Hypothetical Patients With Different Pretest Probabilities
Mesothelin is currently the most studied serum biomarker of malignant pleural mesothelioma. To examine the reported diagnostic accuracies and evaluate whether this biomarker can be an adjunct to an earlier diagnosis of mesothelioma, we performed an IPD meta-analysis on 16 published diagnostic studies, representing a total of 4,491 individuals.
On review, all studies had a good methodologic quality but did show large differences in number of participants, clinical characteristics (age, type of control group, mesothelioma stage, and histologic subtype), and reported diagnostic thresholds of mesothelin. In addition, clinically relevant information concerning mesothelioma stage and histology was often not available. Such heterogeneity cannot be adequately addressed in systematic reviews or meta-analyses based on aggregate mesothelin data.41,42 Conducting an IPD meta-analysis allowed us to adequately quantify and address the observed between-study heterogeneity.
First, we evaluated how the diagnostic accuracy of mesothelin compared across studies. Interestingly, even when the large differences in the applied thresholds were eliminated, the sensitivity and specificity of mesothelin still displayed a substantial between-study heterogeneity. To gain more insight in the sources of this heterogeneity, we performed an ROC regression analysis. Before this analysis, mesothelin levels were standardized to account for the previously reported correlation with age43 and between-laboratory differences.15 Subsequent results showed that the between-study heterogeneity in diagnostic accuracy can be explained by differences in type of control group, mesothelioma stage, and histologic subtype. Mesothelin levels better differentiated patients with mesothelioma from controls without a malignancy than from those with lung cancer, whereas controls were better discriminated from patients with advanced epithelioid or biphasic mesothelioma than from those with early-stage or sarcomatoid disease. These results confirmed the findings of individual studies and are the consequence of mesothelin overexpression in some lung cancers, a lack of mesothelin expression in nonepithelioid mesothelioma, and the association of mesothelin levels with tumor burden.9,12 Altogether, studies that include predominantly young healthy controls and older patients with an epithelioid histology and a more advanced disease are likely to report a high diagnostic accuracy of mesothelin, and vice versa.
To confine such over- or underestimation to a minimum, the use of mesothelin in early diagnosis was evaluated in a clinically relevant model, in which symptomatic or high-risk controls were differentiated from patients with stage I or II epithelioid and biphasic mesothelioma. Although the resulting AUC of mesothelin was acceptable, this value merely provides an indication of its overall diagnostic performance and is of little relevance towards its actual use in clinical practice.20 A mesothelin blood test would especially be useful if it allows clinicians to efficiently steer further diagnostic steps and shorten the current diagnostic delay of mesothelioma. Therefore, we evaluated mesothelin in the following two specific clinical settings: as an adjunct to rule in (through a positive blood test) or to rule out (through a negative blood test) the diagnosis of early-stage mesothelioma. For a negative mesothelin test to aid in excluding mesothelioma, the use of a high-sensitivity threshold (typically in the range of 1.00 to 1.50 nmol/L) is a first requirement. However, our results indicated that at a sensitivity of 95% for mesothelioma, more than 75% of the controls would have false-positive test results, leading to an inordinate number of individuals undergoing unnecessary diagnostic work-ups or biopsies. This is especially relevant for mesothelioma, because this malignancy has a low prevalence, even in populations at risk. As a result, a negative mesothelin test cannot serve as an adjunct in excluding mesothelioma diagnosis, even at a high-sensitivity threshold. For a positive mesothelin test to serve as an adjunct to include diagnosis, the use of a high-specificity threshold (likely in the range of 2.00 to 2.50 nmol/L) is typically required. At a specificity of 95%, however, we found that approximately 70% of patients with early-stage epithelioid or biphasic mesothelioma would remain undetected—an unacceptably high proportion. Although a positive mesothelin test at a high-specificity threshold presents a strong incentive to urge ensuing diagnostic steps (eg, thoracoscopy), its poor sensitivity clearly limits the added value to early diagnosis.
Different approaches can be pursued and combined to anticipate this limited accuracy of serum mesothelin. First, clinicians could use sequential mesothelin measurements to monitor symptomatic patients for marked changes in their blood levels, rather than solely rely on a single mesothelin measurement and a fixed diagnostic threshold.44 When comparing the latter approach with a longitudinal algorithm, a recent retrospective study indeed saw an increase in sensitivity for mesothelioma from 16% to 40%.45 Second, accounting for clinical characteristics that affect serum mesothelin levels, like age, glomerular filtration rate, and body mass index, might also improve the performance of this biomarker.43 Third, the current gold standard for the measurement of serum mesothelin, the Mesomark ELISA, should be critically looked at. In addition to challenging this assay with previously developed mesothelin ELISAs,22,23 the development of more sensitive antibodies is also of interest. Fourth, the quest for more accurate biomarker panels should be pursued. Given the heterogeneity of mesothelioma, it is indeed plausible that a single biomarker cannot provide the necessary sensitivity and specificity for clinical practice. However, none of the studied combinations with mesothelin, including megakaryocyte potentiating factor, osteopontin, carcinoembryonic antigen, CA-125, cytokeratins, and hyaluronic acid, so far managed to substantially outperform mesothelin.16,27,32,33,36 Further biomarker research could therefore specifically focus on patients who lack elevated mesothelin levels. For any candidate biomarker (or combination) of mesothelioma, it will be essential to evaluate its accuracy in direct comparison with mesothelin in a sufficiently large study population including relevant controls and patients with early-stage mesothelioma. Serum mesothelin is currently also under investigation in other fields of mesothelioma management, including monitoring therapy response and estimating patient prognosis.43,4649 It is obvious that further biomarker research is equally relevant for these fields.
Our meta-analysis has some limitations that require consideration. First, IPD was collected from 4,491 individuals, but the ROC regression analyses were performed on smaller groups of participants because of missing data on age, tumor stage, and histologic subtype. Second, cytology was used in a number of studies to diagnose mesothelioma. This approach is typically considered to have a high risk of diagnostic error,2 though experienced centers report reliable results.8 Although the controversy remains, the consequences for our meta-analysis were limited, because only a small number of the patients with mesothelioma (9%) were actually diagnosed with cytology. In addition, these patients lacked data on tumor stage and histology and, therefore, were not included in most of our analyses. Third, we cannot exclude the possibility of a positive publication bias (ie, negative studies on mesothelin that never got published). Fourth, other factors that affect serum mesothelin levels, such as glomerular filtration rate and body mass index,43 were not accounted for, because these were not reported in the original studies. Future studies on mesothelin are strongly encouraged to report all of these clinical characteristics to more efficiently match study participants and interpret the obtained results.
In conclusion, our IPD meta-analysis indentified the presence and the origin of a substantial between-study heterogeneity in the diagnostic accuracy of mesothelin and allowed to evaluate the use of mesothelin in a clinically relevant model. We found that, in symptomatic or high-risk individuals, a negative blood test for mesothelin is not a useful adjunct to exclude mesothelioma, even at a high-sensitivity threshold. Conversely, a positive blood test for mesothelin at a high-specificity threshold was found to be a strong incentive to urge further diagnostic steps and could possibly lead to an earlier diagnosis. However, the associated poor sensitivity of mesothelin for early mesothelioma clearly limits its added value to early diagnosis and emphasizes the need for further biomarker research.
Appendix
Table A1.
Evaluation of the Study Methodology
StudyRepresentative and Well-Defined Controls?Representative and Well-Defined Patient Cases?Blinded Analysis of Serum Mesothelin?Interlaboratory Variance Avoided?
Creaney et al25+++
Scherpereel et al26++++
Creaney et al27+++
Grigoriu et al28++++
Creaney et al29++++
Di Serio et al30++++
Cristaudo et al31++++
van den Heuvel et al32+++
Creaney et al33+++
Amati et al24+++
Pass et al34+++
Schneider et al35++++
Grigoriu et al36++++
Rodriguez Portal et al37+++
Hollevoet et al16++++
Rai et al38+++
Abbreviations: +, yes; −, no.
Table A2.
Studies and Participants Included in the Receiver Operating Characteristic Regression Model to Evaluate the Diagnostic Value of Mesothelin in Early-Stage Mesothelioma
ReferenceControl Groups (No. of participants)
Stage I or II Mesothelioma (No. of participants)
Total No. of Participants
Healthy Asbestos ExposedBenign Asbestos-Related DiseaseBenign Respiratory DiseaseLung CancerOverallBiphasic
Scherpereel et al26; Grigoriu et al28,363233565
Di Serio et al3026661030121144
Cristaudo et al3120312227215363603
van den Heuvel et al32110438153
Pass et al34621743311269
Schneider et al3575139241238
Hollevoet et al16891234663363357
Total31848083731217321,829
Fig A1.
Fig A1.
The origin of serum mesothelin. The mesothelin gene encodes a precursor protein, which is cleaved into the 40-kDa membrane-bound mesothelin and the 31-kDa soluble megakaryocyte potentiating factor (MPF). Mesothelin has three isoforms. Variants 2 and 3 (more ...)
Fig A2.
Fig A2.
Flow chart of the study selection.
Footnotes
K.H. is supported in part by the Foundation Against Cancer, a Belgian foundation of public interest; by the research grant Emmanuel Van der Schueren of the Flemish League Against Cancer, and by the Intramural Research Program of the National Institutes of Health (NIH), National Cancer Institute, Center for Cancer Research. H.I.P. received research support from the Early Detection Research Network of the National Cancer Institute, NIH. B.W.R. and J.C. are supported by the National Health and Medical Research Council of Australia and the Western Australian Insurance Commission.
Presented in part at the 10th International Congress of the International Mesothelioma Interest Group, August 31 to September 3, 2010, Kyoto, Japan.
Study sponsors had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Employment or Leadership Position: None Consultant or Advisory Role: Harvey I. Pass, Fujirebio Diagnostics (C) Stock Ownership: None Honoraria: Harvey I. Pass, Fujirebio Diagnostics Research Funding: Jan P. van Meerbeeck, CIS Bio International; Harvey I. Pass, Fujirebio Diagnostics; Arnaud Scherpereel, CIS Bio International Expert Testimony: None Other Remuneration: None
AUTHOR CONTRIBUTIONS
Conception and design: Kevin Hollevoet, Johannes B. Reitsma, Jan P. van Meerbeeck
Administrative support: Kevin Hollevoet
Provision of study materials or patients: Kevin Hollevoet, Jan P. van Meerbeeck, Paul Baas, Jenette Creaney, Alfonso Cristaudo, Francesca Di Serio, Bogdan D. Grigoriu, Thomas Muley, Kristiaan Nackaerts, Harvey I. Pass, Alex J. Rai, Bruce W. Robinson, José A. Rodríguez Portal, Arnaud Scherpereel, Joachim Schneider, Marco Tomasetti
Collection and assembly of data: Kevin Hollevoet, Jan P. van Meerbeeck, Paul Baas, Jenette Creaney, Alfonso Cristaudo, Francesca Di Serio, Bogdan D. Grigoriu, Thomas Muley, Kristiaan Nackaerts, Harvey I. Pass, Alex J. Rai, Bruce W. Robinson, José A. Rodríguez Portal, Arnaud Scherpereel, Joachim Schneider, Marco Tomasetti
Data analysis and interpretation: Kevin Hollevoet, Johannes B. Reitsma, Jan P. van Meerbeeck, Jenette Creaney, Bogdan D. Grigoriu,Joachim Schneider
Manuscript writing: All authors
Final approval of manuscript: All authors
Affiliations
Kevin Hollevoet and Jan P. van Meerbeeck, Ghent University Hospital, Ghent; Kristiaan Nackaerts, University Hospital Gasthuisberg, Leuven, Belgium; Johannes B. Reitsma, University Medical Center Utrecht, Utrecht; Paul Baas, The Netherlands Cancer Institute, Amsterdam, the Netherlands; Jenette Creaney and Bruce W. Robinson, University of Western Australia, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia; Bogdan D. Grigoriu, University of Medicine, Iasi, Romania; Arnaud Scherpereel, University Hospital (Centre Hospitalier Régional et Universitaire) of Lille II, Lille, France; Alfonso Cristaudo, University of Pisa, Pisa; Francesca Di Serio, University Hospital, Bari; Marco Tomasetti, Polytechnic University of Marche, Ancona, Italy; José A. Rodríguez Portal, Virgen del Rocío University Hospital, Seville, Spain; Joachim Schneider, Justus-Liebig Universität, Giessen; Thomas Muley, Thoraxklinik am Universitätsklinikum Heidelberg, Heidelberg, Germany; Kevin Hollevoet, National Cancer Institute, National Institutes of Health, Bethesda, MD; Harvey I. Pass, New York University, Langone Medical Center and Cancer Center; and Alex J. Rai, Columbia University Medical Center, New York, NY.
1. Robinson BW, Musk AW, Lake RA. Malignant mesothelioma. Lancet. 2005;366:397–408. [PubMed]
2. Scherpereel A, Astoul P, Baas P, et al. Guidelines of the European Respiratory Society and the European Society of Thoracic Surgeons for the management of malignant pleural mesothelioma. Eur Respir J. 2009;35:479–495. [PubMed]
3. LaDou J, Castleman B, Frank A, et al. The case for a global ban on asbestos. Environ Health Perspect. 2010;118:897–901. [PMC free article] [PubMed]
4. Tsao AS, Wistuba I, Roth JA, et al. Malignant pleural mesothelioma. J Clin Oncol. 2009;27:2081–2090. [PubMed]
5. Vogelzang NJ, Rusthoven JJ, Symanowski J, et al. Phase III study of pemetrexed in combination with cisplatin versus cisplatin alone in patients with malignant pleural mesothelioma. J Clin Oncol. 2003;21:2636–2644. [PubMed]
6. van Meerbeeck JP, Gaafar R, Manegold C, et al. Randomized phase III study of cisplatin with or without raltitrexed in patients with malignant pleural mesothelioma: An intergroup study of the European Organisation for Research and Treatment of Cancer Lung Cancer Group and the National Cancer Institute of Canada. J Clin Oncol. 2005;23:6881–6889. [PubMed]
7. Weder W, Opitz I, Stahel R. Multimodality strategies in malignant pleural mesothelioma. Semin Thorac Cardiovasc Surg. 2009;21:172–176. [PubMed]
8. Whitaker D. The cytology of malignant mesothelioma. Cytopathology. 2000;11:139–151. [PubMed]
9. Hassan R, Ho M. Mesothelin targeted cancer immunotherapy. Eur J Cancer. 2008;44:46–53. [PMC free article] [PubMed]
10. Scholler N, Fu N, Yang Y, et al. Soluble member(s) of the mesothelin/megakaryocyte potentiating factor family are detectable in sera from patients with ovarian carcinoma. Proc Natl Acad Sci U S A. 1999;96:11531–11536. [PubMed]
11. Hellstrom I, Raycraft J, Kanan S, et al. Mesothelin variant 1 is released from tumor cells as a diagnostic marker. Cancer Epidemiol Biomarkers Prev. 2006;15:1014–1020. [PubMed]
12. Robinson BW, Creaney J, Lake R, et al. Mesothelin-family proteins and diagnosis of mesothelioma. Lancet. 2003;362:1612–1616. [PubMed]
13. Li Z, Verch T, Allard JW. MESOMARK in vitro diagnostic test for mesothelioma. Expert Opin Med Diagn. 2007;1:137–142. [PubMed]
14. Whiting P, Rutjes AW, Reitsma JB, et al. The development of QUADAS: A tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol. 2003;3:25. [PMC free article] [PubMed]
15. Beyer HL, Geschwindt RD, Glover CL, et al. MESOMARK: A potential test for malignant pleural mesothelioma. Clin Chem. 2007;53:666–672. [PubMed]
16. Hollevoet K, Nackaerts K, Thimpont J, et al. Diagnostic performance of soluble mesothelin and megakaryocyte potentiating factor in mesothelioma. Am J Respir Crit Care Med. 2010;181:620–625. [PubMed]
17. Reitsma JB, Glas AS, Rutjes AW, et al. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol. 2005;58:982–990. [PubMed]
18. Riley RD, Higgins JP, Deeks JJ. Interpretation of random effects meta-analyses. BMJ. 2011;342:964–967. [PubMed]
19. Janes H, Pepe MS. Adjusting for covariates in studies of diagnostic, screening, or prognostic markers: An old concept in a new setting. Am J Epidemiol. 2008;168:89–97. [PubMed]
20. Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010;5:1315–1316. [PubMed]
21. Grimes DA, Schulz KF. Refining clinical diagnosis with likelihood ratios. Lancet. 2005;365:1500–1505. [PubMed]
22. Hassan R, Remaley AT, Sampson ML, et al. Detection and quantitation of serum mesothelin, a tumor marker for patients with mesothelioma and ovarian cancer. Clin Cancer Res. 2006;12:447–453. [PubMed]
23. Iwahori K, Osaki T, Serada S, et al. Megakaryocyte potentiating factor as a tumor marker of malignant pleural mesothelioma: Evaluation in comparison with mesothelin. Lung Cancer. 2008;62:45–54. [PubMed]
24. Amati M, Tomasetti M, Scartozzi M, et al. Profiling tumor-associated markers for early detection of malignant mesothelioma: An epidemiologic study. Cancer Epidemiol Biomarkers Prev. 2008;17:163–170. [PubMed]
25. Creaney J, Christansen H, Lake R, et al. Soluble mesothelin related protein in mesothelioma. J Thorac Oncol. 2006;1:172–174. [PubMed]
26. Scherpereel A, Grigoriu B, Conti M, et al. Soluble mesothelin-related peptides in the diagnosis of malignant pleural mesothelioma. Am J Respir Crit Care Med. 2006;173:1155–1160. [PubMed]
27. Creaney J, van Bruggen I, Hof M, et al. Combined CA125 and mesothelin levels for the diagnosis of malignant mesothelioma. Chest. 2007;132:1239–1246. [PubMed]
28. Grigoriu BD, Scherpereel A, Devos P, et al. Utility of osteopontin and serum mesothelin in malignant pleural mesothelioma diagnosis and prognosis assessment. Clin Cancer Res. 2007;13:2928–2935. [PubMed]
29. Creaney J, Yeoman D, Naumoff LK, et al. Soluble mesothelin in effusions: A useful tool for the diagnosis of malignant mesothelioma. Thorax. 2007;62:569–576. [PMC free article] [PubMed]
30. Di Serio F, Fontana A, Loizzi M, et al. Mesothelin family proteins and diagnosis of mesothelioma: Analytical evaluation of an automated immunoassay and preliminary clinical results. Clin Chem Lab Med. 2007;45:634–638. [PubMed]
31. Cristaudo A, Foddis R, Vivaldi A, et al. Clinical significance of serum mesothelin in patients with mesothelioma and lung cancer. Clin Cancer Res. 2007;13:5076–5081. [PubMed]
32. van den Heuvel MM, Korse CM, Bonfrer JM, et al. Non-invasive diagnosis of pleural malignancies: The role of tumour markers. Lung Cancer. 2008;59:350–354. [PubMed]
33. Creaney J, Yeoman D, Demelker Y, et al. Comparison of osteopontin, megakaryocyte potentiating factor, and mesothelin proteins as markers in the serum of patients with malignant mesothelioma. J Thorac Oncol. 2008;3:851–857. [PubMed]
34. Pass HI, Wali A, Tang N, et al. Soluble mesothelin-related peptide level elevation in mesothelioma serum and pleural effusions. Ann Thorac Surg. 2008;85:265–272. [PubMed]
35. Schneider J, Hoffmann H, Dienemann H, et al. Diagnostic and prognostic value of soluble mesothelin-related proteins in patients with malignant pleural mesothelioma in comparison with benign asbestosis and lung cancer. J Thorac Oncol. 2008;3:1317–1324. [PubMed]
36. Grigoriu B, Chahine B, Zerimech F, et al. Serum mesothelin has a higher diagnostic utility than hyaluronic acid in malignant mesothelioma. Clin Biochem. 2009;42:1046–1050. [PubMed]
37. Rodriguez Portal JA, Rodriguez Becerra E, Rodriguez Rodriguez D, et al. Serum levels of soluble mesothelin-related peptides in malignant and nonmalignant asbestos-related pleural disease: Relation with past asbestos exposure. Cancer Epidemiol Biomarkers Prev. 2009;18:646–650. [PubMed]
38. Rai AJ, Flores RM, Mathew A, et al. Soluble mesothelin related peptides (SMRP) and osteopontin as protein biomarkers for malignant mesothelioma: Analytical validation of ELISA based assays and characterization at mRNA and protein levels. Clin Chem Lab Med. 2010;48:271–278. [PubMed]
39. Rusch VW. A proposed new international TNM staging system for malignant pleural mesothelioma from the International Mesothelioma Interest Group. Lung Cancer. 1996;14:1–12. [PubMed]
40. American Joint Committee on Cancer. Handbook for Staging of Cancer. ed 4. Philadelphia, PA: J.B. Lippincott Co; 1993. pp. 137–138.
41. Luo L, Shi HZ, Liang QL, et al. Diagnostic value of soluble mesothelin-related peptides for malignant mesothelioma: A meta-analysis. Respir Med. 2009;104:149–156. [PubMed]
42. van der Bij S, Schaake E, Koffijberg H, et al. Markers for the non-invasive diagnosis of mesothelioma: A systematic review. Br J Cancer. 2011;104:1325–1333. [PMC free article] [PubMed]
43. Hollevoet K, Nackaerts K, Thas O, et al. The effect of clinical covariates on the diagnostic and prognostic value of soluble mesothelin and megakaryocyte potentiating factor. Chest. [epub ahead of print on July 7, 2011] [PubMed]
44. Hollevoet K, Van Cleemput J, Thimpont J, et al. Serial measurements of mesothelioma serum biomarkers in asbestos-exposed individuals: A prospective longitudinal cohort study. J Thorac Oncol. 2011;6:889–895. [PubMed]
45. Creaney J, Olsen NJ, Brims F, et al. Serum mesothelin for early detection of asbestos-induced cancer malignant mesothelioma. Cancer Epidemiol Biomarkers Prev. 2010;19:2238–2246. [PubMed]
46. Grigoriu BD, Chahine B, Vachani A, et al. Kinetics of soluble mesothelin in patients with malignant pleural mesothelioma during treatment. Am J Respir Crit Care Med. 2009;179:950–954. [PubMed]
47. Wheatley-Price P, Yang B, Patsios D, et al. Soluble mesothelin-related peptide and osteopontin as markers of response in malignant mesothelioma. J Clin Oncol. 2010;28:3316–3322. [PubMed]
48. Creaney J, Francis RJ, Dick IM, et al. Serum soluble mesothelin concentrations in malignant pleural mesothelioma: Relationship to tumor volume, clinical stage and changes in tumor burden. Clin Cancer Res. 2011;17:1181–1189. [PubMed]
49. Hollevoet K, Nackaerts K, Gosselin R, et al. Soluble mesothelin, megakaryocyte potentiating factor, and osteopontin as markers of patient response and outcome in mesothelioma. J Thorac Oncol. 2011;6:1930–1937. [PubMed]
Articles from Journal of Clinical Oncology are provided here courtesy of
American Society of Clinical Oncology