Lung Cancer (2008) 59, 350 354 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/lungcan Non-invasive diagnosis of pleural malignancies: The role of tumour markers Michel M. van den Heuvel a, Catharina M. Korse b, Johannes M.G. Bonfrer b, Paul Baas a, a Department of Thoracic Oncology, The Netherlands Cancer Institute-Antoni van Leeuwenhoek Hospital, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands b Department of Clinical Chemistry, The Netherlands Cancer Institute-Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands Received 16 May 2007; received in revised form 6 July 2007; accepted 27 August 2007 KEYWORDS Mesothelioma; Non-small cell lung cancer; Pleural effusion; Soluble mesothelin-related protein; CEA; Cyfra 21.1 Summary Background: Serum markers have been tested in patients with malignant effusions for their ability to differentiate malignant pleural mesothelioma from other causes. We have assessed three different serum markers and report our findings in a series of patients with different thoracic malignancies and a group of healthy controls. Methods: A retrospective analysis of 179 patients and 50 healthy controls was performed. Seventy-four patients had a confirmed mesothelioma, and 106 patients had non-small cell lung cancer (NSCLC), 55 had adeno. Soluble mesothelin-related protein (SMRP), Cyfra 21.1, and carcino-embrionic antigen (CEA) were tested in serum at time of presentation. Results: Cyfra 21.1 was the best single marker discriminating healthy from any malignant disease (area under receiver operating characteristics curve (AUC): 0.92; 95% confidence interval (CI): 0.89 0.96). By combining all three markers the discriminatory power improved marginally (AUC: 0.95; CI: 0.93 0.98; p = 0.015). The combination of CEA and SMRP was most accurate in differentiating mesothelioma from NSCLC (AUC: 0.94; 95% CI: 0.90 0.97) and could correctly identify 152 of 179 (85%) cases. Conclusions: By using two serum markers (CEA and SMRP) we were able to discriminate mesothelioma from NSCLC with high sensitivity, while Cyfra 21.1 is useful in the discrimination of normal versus malignancy. 2007 Elsevier Ireland Ltd. All rights reserved. 1. Introduction Corresponding author. Tel.: +31 20 512 2958; fax: +31 20 512 2572. E-mail address: p.baas@nki.nl (P. Baas). The last decade has shown a rise in the incidence of malignant pleural mesothelioma (MPM) which is expected to peak between 2015 and 2020 in the developed world. 0169-5002/$ see front matter 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2007.08.030
The role of tumour markers 351 MPM is closely related with asbestos exposure and many countries have now accepted legal issues for financial reimbursement for patients who have had an occupational asbestos exposure. The increased incidence and availability of more effective treatment options, have lead to a greater awareness in both patients and doctors to obtain a histopathological confirmation of the disease [1]. In general, cytological examination will be the first step to discriminate between a malignant or benign origin in patients presenting with a pleural effusion or mass. Cytology is quite effective in diagnosing pleural metastasis (yield of 60 90%, depending on the extend and type of malignancy) but can identify MPM in only a minority of the cases (average yield of 30 50%) [2]. Differentiation between adeno and mesothelioma can be cumbersome [3,4] and cytology is not always accepted as a prove for financial reimbursement. More invasive procedures such as blind or open pleural biopsies, medical or video assisted thoracoscopy are frequently required to establish a diagnosis [5]. It would be helpful to have methods to speed up the diagnostic process and identify patients who require invasive procedures at an early time. Tumour makers could be considered to be of help in this setting. Cyfra 21.1 is an assay that determines the presence of cytokeratin factors that are shed in the plasma during cell division and decay. These cytokeratin fragments are intracytoplasmatic filaments which are expressed by epithelial and mesothelial cells, and tumours of these tissues. Increased levels can be found in patients with a variety of malignancies like lung, breast, pancreatic cancer and MPM [6]. It has been shown that these markers can be used to monitor treatment response and that they carry a prognostic value in MPM [7]. Carcino-embrionic antigen (CEA) is an oncofetal glycoprotein that is used to confirm the diagnosis and for monitoring of treatment effects. In pleural malignancies it has been used to differentiate mesothelioma from other causes [8 10]. Recent publications on the use of soluble mesothelin-related protein (SMRP) have attracted attention for their superior sensitivity and specificity in mesothelioma [11,12]. Mesothelin is a member of the SMRP family of proteins and has been identified in pancreatic cancer, ovarian cancer and MPM. The value of SMRP in differentiating malignant pleural diseases has yet to be established. With the new promising SMRP test we decided to evaluate whether a combination of serum tumour markers (CEA, SMRP, and Cyfra 21.1) would help us to make a distinction between the different types of thoracic malignancies. Patients were compared to healthy individuals in order to identify the optimal cut off values of the tumour markers. Subsequently, patients were grouped according to their histology in order to assess the predictive value of the combination of serum tumour markers. 2. Materials and methods 2.1. Patients and controls Serum samples of patients presenting at the Netherlands Cancer Institute, a national cancer centre, with pleural effusions or pleural masses were stored in 30 C until analysis. All available serum samples of MPM patients presented in the period between 1994 and 2004 were included. In addition, a random selection of 106 patients with NSCLC and 50 healthy controls was performed. In all patients with pleural metastasis a histological diagnosis was obtained. In case of MPM, the diagnosis and subtype was confirmed by a national expert panel. Healthy control subjects were recruited among companions of patients visiting the outpatient department. All persons gave informed consent for analysis of serum tumour markers. CEA and Cyfra 21.1 were measured on the E170 analyser, which is based on chemiluminescent immunometric technology (Roche Diagnostics, Mannheim, Germany). SMRP concentrations were determined using a double determinant enzyme linked immunosorbent assay using. The kits were kindly provided by Cisbio (Toulouse, France). 2.2. Statistics Upper reference values were defined by the geometric mean + twice standard deviation of the healthy control group. Student s t-test was used to compare the mean age between the groups. Univariate and multiple logistic regressions were performed to predict the presence of a disease. Linear relationship between the tumour markers and the logit of the outcome variable was investigated. Receiver operating characteristic (ROC) curves were used to compare the diagnostic performance of the tumour markers [13,14]. Inverse values were represented when appropriate. 3. Results Serum samples were available of 50 volunteers and 179 patients with malignant pleural disease (Table 1). Non-small cell lung (NSCLC) was diagnosed in 106 patients and MPM in 73. The majority of patients with squamous cell and mesothelioma patients were male, whereas in the other groups this male/female distribution was well balanced. The mean age of the healthy controls was 12 years younger than patients with malignant pleural disease (p < 0.001). Table 1 Patient characteristics Group Number Age (years) Male (%) Healthy controls 50 48 ± 10 50 Malign Pleural 73 61 ± 8 99 Mesothelioma Epithelial 43 60 ± 8 100 Mixed 15 63 ± 8 100 Sarcomatous 10 60 ± 7 100 NOS 5 60 ± 14 80 Non-small cell 106 59 ± 11 64 lung Adeno 55 56 ± 12 54 Squamous cell 35 60 ± 11 86 Large cell, other 16 58 ± 12 50 NOS, not otherwise specified.
352 M.M. van den Heuvel et al. Table 2 Percentage of patients with increased levels of tumour markers Group Number CEA Cyfra 21.1 SMRP Malign pleural 73 1 66 60 mesothelioma Epithelial 43 2 72 67 Mixed 15 0 67 40 Sarcomatous 10 0 40 60 NOS 5 0 60 60 Non-small cell lung 106 52 67 21 Adeno 55 75 56 22 Squamous cell 35 20 71 17 Large cell, other 16 44 69 25 NOS, not otherwise specified. The geometric mean SMRP concentration in the healthy group of 50 individuals was 0.2 nmol/l and the upper reference value was determined at 1.3 nmol/l in this population. Data for CEA were 2 and 6 g/l, respectively. The Cyfra 21.1 geometric mean was 0.6 and the upper reference value 1.9 g/l. In the group of 73 patients with MPM, 60% of the SMRP values were above the reference level comparable to the 66% elevated Cyfra 21.1 values. In contrast, only 1 MPM patient had an increased CEA level (1%) at a value of 8 g/l. In the 106 patients with NSCLC, SMRP was increased in 21%, Cyfra 21.1 in 63% and CEA in 52% of the sera (Table 2). SMRP and Cyfra 21.1 levels were increased in comparable figures in the various histological types of NSCLC, while CEA was elevated in a high number of adenos (75%) and only in 20% of the cases with squamous cell. The distribution of tumour marker serum concentrations was given by drawing box plots (Fig. 1). No significant correlations existed between the three different markers (data not shown). Serum SMRP, Cyfra 21.1, and CEA were shown to be independent predictors of thoracic malignancy (Table 3). A ROC curve illustrated that discerning the patients with malignancy from the healthy group is best done by using a combination of all three markers (AUC: 0.95; CI: 0.93 0.98) (Fig. 2). The AUC of this combination represented a small, Table 3 Results of logistic regression models that predict the presence of a thoracic malignancy by ways of serum tumour markers Tumour marker OR 95% CI p-value Univariate analysis Cyfra 21.1 ( g/l) 21.19 7.15 62.80 <0.001 SMRP (nmol/l) 6.50 2.56 16.52 <0.001 CEA ( g/l) 1.19 1.03 1.36 0.016 Multiple analysis Cyfra 21.1 ( g/l) 13.74 4.62 40.85 <0001 SMRP (nmol/l) 6.62 1.83 23.93 0.004 CEA ( g/l) 1.24 1.02 1.50 0.027 Fig. 1 Distribution of the tumour markers in healthy persons, patients with non-small cell lung (NSCLC) and patients with malignant pleural mesothelioma (MPM). Boxplots shows the median, interquartile range and outliers. The dotted lines represent the upper reference value. but significant improvement compared to Cyfra 21.1 alone (AUC: 0.92; CI: 0.89 0.96; p = 0.015). SMRP and CEA were helpful to discriminate MPM from NSCLC, whereas Cyfra 21.1 had no predictive value. In the multiple analyses SMRP and CEA were independent predictors of the diagnosis (Table 4). The predictive value of the SMRP/CEA combination was 85% for all 179 cases. To visualize the optimal use of SMRP, CEA and Cyfra 21.1 we constructed a ROC curve including the three markers and the combination of SMRP and CEA (Fig. 3). The AUC of the
The role of tumour markers 353 Fig. 2 Receiver operating characteristics (ROC) curves of CEA ( ), Cyfra21.1 ( ), SMRP ( ) and the combination of the three markers ( ) from healthy persons and patients with non-small cell lung (NSCLC) or malignant pleural mesothelioma (MPM). combination (0.94; CI: 0.90 0.97) was a significant improvement compared to the inverse value of CEA alone (0.90; CI: 0.85 0.94). At a specificity of 95% a sensitivity of 66% was achieved. 4. Discussion In pleural diseases it remains a challenge to obtain a clear cut diagnosis. A number of tumour markers has shown to facilitate the diagnostic process in patients with an assumed pleural malignancy [15,16]. With the current retrospective study we provide data that such an approach, using a set of serum tumour markers, can potentially prevent further diagnostic interventions in a proportion of patients with a malignant pleural effusion. Research on diagnostic techniques of suspected malignant pleural disease has focused on the question whether lab measurements could replace invasive and time consuming procedures. The measurement of tumour products or Table 4 Results of logistic regression models to distinguish malignant pleural mesothelioma from malignant lung disease with the tumour markers as predictor variables Tumour marker OR 95% CI p-value Univariate analysis Cyfra 21.1 ( g/l) 1.00 0.99 1.02 0.515 SMRP (nmol/l) 1.64 1.26 2.13 <0.001 CEA ( g/l) 0.45 0.33 0.61 <0.001 Multiple analysis SMRP (nmol/l) 2.02 1.27 3.21 0.003 CEA ( g/l) 0.43 0.30 0.62 <0.001 Fig. 3 Receiver operating characteristics (ROC) curves of the inverse levels of CEA ( ), Cyfra21.1 ( ), SMRP ( ) and the combination of CEA and SMRP ( ) from patients with non-small cell lung (NSCLC) and malignant pleural mesothelioma (MPM). substances in serum and pleural fluid could be a solution to this question [16,17]. Diagnostic algorithms were developed based on tumour marker measurements in pleural effusion [16]. For instance, high levels of CEA, CA15.3, and Cyfra21.1 are now considered indicative for more invasive diagnostic procedures to determine the presence of malignant disease whereas low levels suggest benign disease. Poor test accuracy and important intra-study variance however, hamper the introduction of these types of algorithms in clinical routine. Besides, these algorithms have never been tested prospectively. In this study we tested the accuracy of a panel of tumour markers in patients who presented with a malignancy. Using this panel of markers we were able to differentiate between patients with MPM, NSCLC, and healthy individuals. This study focused on two parts. Firstly, our aim was to identify which marker, or combination of markers, was best in differentiating healthy controls from the patients with a malignant disease. Despite the relatively poor test accuracy Cyfra 21.1 was the best single predictor. The test performance was improved by using a combination of all three markers but these differences are fairly small. The second analysis was performed to determine the potential of these tumour markers to discriminate between MPM and NSCLC. By combining serum CEA and SMRP we were able to distinguishing NSCLC from mesothelioma more accurately and by its use pathological examination could potentially be avoided in the majority of cases. It is well known that CEA is a negative predictor for MPM [8,9]. In the last few years, other markers such as SMRP have been introduced that identify MPM. The value of SMRP in serum and pleural fluid in diagnosing MPM was investigated retrospectively by Robinson et al. and they were able to identify patients at risk for the development of
354 M.M. van den Heuvel et al. mesothelioma [12,18]. The sensitivity however was not high enough and therefore the authors stated that histo- or cytopathological examination of pleural lesions remained the cornerstone of diagnosis. Scherpereel et al. [11] prospectively investigated the value of SMRP in patients with MPM, other pleural malignancies and benign pleural disorders. For SMRP the AUC values of the ROC curves were significantly higher in patients with MPM compared to controls with benign disease (sensitivity 80%, specificity 82%), while differentiation of MPM from other malignant diseases was significantly lower (sensitivity 58%, specificity 73%). The serum SMRP data in the current study seem to be compatible to the data of Scherpereel. One of the other interesting tumour markers tested in MPM is osteopontin. This marker was tested by Pass et al. who differentiated MPM patients from asbestos exposed persons with a sensitivity of 78% and a specificity of 86% [19]. However, a recent publication showed that it adds no value to distinguish MPM from adeno [20]. Wecan summarize that single tumour markers remain of limited diagnostic value. With the use of combinations of different markers the test accuracy improves to some extent and it has become feasible by using serum tumour markers to differentiate between the various causes of pleural effusion [16,17]. The presented results confirm earlier studies that used different sets of tumour markers. The limited accuracy of individual tumour markers will make it difficult to determine the panel upfront. Our study shows that especially CEA and SMRP are useful in this setting and Cyfra 21.2 may have additional value. However, other known markers such as osteopontin, CA 125 and CA 15-3 also have to be tested to optimise the diagnostic potential of such a panel. The required number of markers for the panel types will have to be assessed in properly designed prospective studies. Conflict of interest statement None. References [1] Baas P, Hullenaar N, Wagenaar J, Kaajan JPG, Koolen M, Schrijver M, et al. Occupational asbestos exposure: how to deal with suspected mesothelioma cases: the Dutch approach. Ann Oncol 2006;17:848 52. [2] Antony VB, Loddenkemper R, Astoul P, Boutin C, Goldstraw P, Hott J, et al. Management of malignant pleural effusions. Eur Respir J 2001;18:402 19. [3] Loddenkemper R. Thoracoscopy state of the art. Eur Respir J 1998;11:213 21. [4] Light RW, Errozan YS, Ball Jr WC. Cells in pleural fluid. Their value in differential diagnosis. Arch Int Med 1973:854 60. [5] Maskell NA, Butland RJA. BTS guidelines for the investigation of a unilateral pleural effusion in adults. Thorax 2003;58:8ii 17ii. [6] Barak V, Goike H, Panaretakis K, Einarsson R. Clinical utility of cytokeratins as tumor markers. Clin Biochem 2004;37:529 40. [7] Schouwink H, Korse C, Bonfrer J, Hart A, Baas P. Prognostic value of the serum tumour markers cyfra 21-1 and tissue polypeptide antigen in malignant mesothelioma. Lung Cancer 1999;25:25 32. [8] Faravelli B, D Amore E, Nosenzo M, Betta PG, Donna A. Carcinoembryonic antigen in pleural effusions. Diagnostic value in malignant mesothelioma. Cancer 1984;5:1194 7. [9] Fuhrman C, Duche JC, Chouaid C, Abd Alsamad I, Atassi K, Monnet I, et al. Use of tumor markers for differential diagnosis of mesothelioma and secondary pleural malignancies. Clin Biochem 2000;33:405 10. [10] Paganuzzi M, Onetto M, Marroni P, Filiberti R, Tassara E, Parodi S, et al. Diagnostic value of Cyfra 21-1 tumor marker and CEA in pleural effusion due to mesothelioma. Chest 2001;119:1138 42. [11] Scherpereel A, Grigoriu B, Conti M, Gey T, Gregoire M, Copin M, et al. Soluble mesothelin-related peptides in the diagnosis of malignant pleural mesothelioma. Am J Respir Crit Care Med 2006;173:1155 60. [12] Robinson B, Creaney J, Lake R, Nowak A, Musk AW, De Klerk N, et al. Mesothelin-family proteins and diagnosis of mesothelioma. Lancet 2003;362:1612 6. [13] Griner PF, Mayewski RJ, Mushlin AI, Greenland P. Selection and interpretation of diagnostic tests and procedures. Principles and applications. Ann Int Med 1981;94:557 92. [14] Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:283 98. [15] Porcel J, Vives M, Esquerda A, Salud A, Perez B, Rodriguez- Panadero F. Use of a panel of tumor markers (carcinoembryonic antigen, cancer antigen 125, carbohydrate antigen 15-3, and cytokeratin 19 fragments) in pleural fluid for the differential diagnosis of benign and malignant effusions. Chest 2004;126:1757 63. [16] Shitrit D, Zingerman B, Shitrit A, Shlomi D, Kramer M. Diagnostic value of Cyfra 21-1, CEA, CA 19-9, CA 15-3, and CA 125 assays in pleural effusions: analysis of 116 cases and review of the literature. Oncologist 2005;10:501 7. [17] Alatas F, Alatas O, Metintas M, Colak O, Harmanci E, Demir S. Diagnostic value of CEA, CA 15-3, CA 19-9, Cyfra 21-1, NSE and TSA assay in pleural effusions. Lung Cancer 2001;31:9 16. [18] Robinson BW, Creaney J, Lake R, Nowak A, Musk AW, De Klerk N, et al. Soluble mesothelin-related protein: a blood test for mesothelioma. Lung Cancer 2005;49(Suppl. 1):S109 11. [19] Pass HI, Lott D, Lonardo F, Harbut M, Liu Z, Tang N, et al. Asbestos exposure, pleural mesothelioma, and serum osteopontin levels. N Engl J Med 2005;353:1564 73. [20] Grigoriu BD, Scherpereel A, Devos P, Chahine B, Letourneux M, Lebailly P, et al. Utility of osteopontin and serum mesothelin in malignant pleural mesothelioma diagnosis and prognosis assessment. Clin Cancer Res 2007;13:2928 35.