Evaluation of a Panel of Molecular Markers for the Diagnosis of Malignant Serous Effusions

Imaging, Diagnosis, Prognosis

Evaluation of a Panel of Molecular Markers for the Diagnosis of Malignant Serous Effusions Karine Passebosc-Faure,1Guorong Li,1Claude Lambert,1Miche`le Cottier,3 Anne Gentil-Perret,4 Pierre Fournel,2 Maurice Pe´rol,5 and Christian Genin1

Abstract Purpose: Our main goal was to evaluate a panel of molecular markers for the detection of cancer cells in serous effusions and to determine their value as an adjunctive reverse transcription-PCR (RT-PCR) test to cytologic examination. Experimental Design: One hundred fourteen serous effusions from 71patients with tumors and 43 patients with benign diseases were subjected to RT-PCR for expression of carcinoembryonic antigen (CEA), epithelial cell adhesion molecule (Ep-CAM), E-cadherin, mammaglobin, mucin 1 (MUC1)isoforms MUC1/REP, MUC1/Y,and MUC1/Z, calretinin, and Wilms’ tumor1susceptibility gene. Results: CEA, Ep-CAM, E-cadherin, and mammaglobin were specifically expressed in malignant effusions. The sensitivity of RT-PCR in cytologically negative malignant effusions was 63.1% combining CEA and Ep-CAM (with 100% specificity) and reached 78.9% adding MUC1/Y or MUC1/Z (with 93% specificity). In the whole population of effusions, the combination of cytology with RT-PCR of CEA and Ep-CAM yielded a 90.1% sensitivity, a specificity and a positive predictive value of 100%, and a 86% negative predictive value for malignancy. Adding MUC1/Yor MUC1/Z to the panel, the sensitivity was 94.5% with 93% specificity, 95.7% PPV, and 90.9% negative predictive value. Moreover, CEA and mammaglobin were specifically expressed in epithelial malignancies, and mammaglobin was mainly expressed in effusions from breast carcinoma (97.3% of specificity). Conclusions: A combination of cytology and RT-PCR analysis of CEA and Ep-CAM significantly improved the detection sensitivity of tumor cells in serous effusions. RT-PCR analysis of CEA, Ep-CAM, and mammaglobin in serous effusions could be a beneficial adjunct to cytology for the diagnosis of malignancy.

Serous effusions are commonly found in patients with cancer matory cells (2). Then, when cancer cells are found in the fluid, and they sometimes reveal the malignancy (1). As the the second question is to know if they are local or metastatic diagnosis of the effusion acts directly on the prognosis and cancer cells, and if they are metastatic, from which organ? the treatment of the disease (the presence of tumor cells in Cytology, which is the gold standard method for the diagnosis effusions point out a poor prognosis and often implies an of serous effusion, has some difficulties to distinguish aggressive treatment, such as chemotherapy), it is clearly mesothelioma cells from carcinoma cells (2). Thus, many important to detect any cancer cells in the fluid with the additional methods have been evaluated (3, 4) to improve the greatest specificity and sensitivity. However, the search for diagnostic accuracy to avoid invasive diagnostic techniques, cancer cells in serous effusions using classic cytologic technique such as thoracoscopy. is often difficult either because of their scarceness or because of Recently, PCR techniques have been largely evaluated for their delicate distinction with reactive mesothelial or inflam- the detection of cancer cells in blood, bone, and lymph nodes and they have proven to be more sensitive than conventional techniques (5, 6). However, PCR was less

Authors’ Affiliations: Departments of 1Immunology, 2Pneumology, 3Cytology, intensively used in serous effusions, although it provides an 4Pathology, University Hospital Center, Saint Etienne, France and 5Department of excellent material for molecular analysis. The few studies that Pneumology, University Hospital Center, Lyon, France have used PCR techniques for the detection of cancer cells in Received 1/7/05; revised 7/7/05; accepted 7/13/05. serous effusions have concluded it to be a beneficial adjunct Grant support: Ligue Contre le Cancer, Comite´ de la Loire, and French Ministry of to conventional techniques because they generally increase Health. The costs of publication of this article were defrayed in part by the payment of page the detection sensitivity of tumor cells (7–11). However, the charges. This article must therefore be hereby marked advertisement in accordance validated markers are very limited, which hinders clinical with 18 U.S.C. Section 1734 solely to indicate this fact. application. Requests for reprints: Christian Genin, Department of Immunology, Bellevue The aims of our study were (a) to evaluate a panel of University Hospital Center, 42055 Saint Etienne, France. Phone: 33-4-77-82- 83-79; Fax: 33-4-77-82-84-93; E-mail: [email protected]. molecular markers for the detection of cancer cells in serous F 2005 American Association for Cancer Research. effusions, (b) to determine the value of the markers as an doi:10.1158/1078-0432.CCR-05-0043 adjunctive reverse transcription-PCR (RT-PCR) test to cytologic

Clin Cancer Res 2005;11(19) October 1, 2005 6862 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Molecular Markers for Cancer Cells in Serous Effusions

Ta b l e 1. Expression of the different markers in serous effusions by RT-PCR

Diagnosis No. CEA Ep-CAM CDH1 MUC1/ YZ* CALB2 MGB1 effusions positive (%) positive (%) positive (%) positive (%) positive (%) positive (%) c Benign 43 0 (0) 0 (0) 0 (0) 3 (7) 1(2.3) 0(0) b c Carcinomas 51 28 (54.9) 38 (74.5) 26 (51) 24 (47) 4(7.8)x 11 (21.6)k Breast 14 6 (42.8) 10 (71.4) 6 (42.8) 4 (28.6) 2 (14.3) 10 (71.4)k Lung{ 12 9(75) 10(83.3) 9(75) 8(66.7) 1(8.33) 0(0) Ovary 3 0 (0) 3 (100) 2 (66.7) 3 (100) 0 (0) 1 (33.3) Others** 13 7 (53.8) 9 (69.2) 6 (46.1) 7 (53.8) 0 (0) 0 (0) Unknown primary site 9 6 (66.7) 6 (66.7) 3 (33.3) 2 (22.2) 1 (11.1) 0 (0) b c Nonepithelial malignancies 20 0 (0) 9 (45) 7(35) 9(45) 5 (25) 0 (0) Mesothelioma 14 0 (0) 8 (57.1) 6 (42.8) 6 (42.8) 5 (35.7)x 0(0) Sarcoma 3 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Lymphoma 2 0 (0) 0 (0) 0 (0) 2 (100) 0 (0) 0 (0) Melanoma 1 0 (0) 1 (100) 1 (100) 1 (100) 0 (0) 0 (0)

*MUC1/Yand MUC1/Z were always coexpressed. cSignificantlymoreexpressedinmalignanteffusions(P < 0.0001). bSignificantly more expressed in carcinomas (P =0.018). xSignificantly more expressed in mesothelioma (P =0.018). kSignificantly more expressed in breast carcinomas (P < 0.0001). {Including nine adenocarcinoma and three small cell lung carcinoma. **Nasopharyngeal (n = 2), stomach (n = 2), pancreas (n =2),kidney(n = 2), uterus (n =1),colon(n = 1), rectum (n = 1), bladder (n = 1), and prostate (n =1).

examination, (c) and to assess their usefulness as differential negative. The cancer primary sites of the 71 patients with malignant diagnostic markers between epithelial and nonepithelial malig- serous effusions are detailed in Table 1. This study was approved by the nant effusions. institutional ethics committee. The markers have been chosen based on their published Preparation of cells from serous effusions. Each fluid sample was centrifuged in 50 mL tubes at 300 g for 10 minutes at room interest for tumor cell detection in serous effusions: mucin 1 temperature. Then, pellets were pooled together, washed once, and (MUC1; designed as MUC1/REP; ref. 9), E-cadherin (CDH1; resuspended in RPMI 1640 with 10% FCS. Fractions of 900 AL were ref. 12), epithelial cell adhesion molecule (Ep-CAM or distributed in 1.5 mL cryotubes containing 100 AL DMSO and stored in TACSTD1; refs. 10, 13), and carcinoembryonic antigen (CEA liquid nitrogen until use. When needed, aliquots were thawed in a 37jC or CEACAM5; ref. 14). We also studied the expression of the water bath and then washed once with Ca2+/Mg2+ –free NaCl solution two isoforms of MUC1, MUC1/Y and MUC1/Z, because they (0.01 mol/L), and a cell count was done. have been found to be preferentially expressed in epithelial RNA extraction. Total RNA was purified using the RNABle extraction tumors compared with MUC1/REP (15). Wilms’ tumor 1 solution (Eurobio, Les Ulis, France) following the manufacturer’s 6 susceptibility gene (WT1; ref. 16) and calretinin (CALB2; refs. instructions using 0.2 mL RNABle per 10 cells. Briefly, cell pellets were 12, 17), two markers of mesothelial cells, were tested to homogenized with the RNABle solution (containing guanidium distinguish mesothelioma from carcinoma effusions. Mamma- isothiocyanate and phenol); then, total RNA was extracted by adding 0.1 volume of chloroform and precipitated using 1 volume of iso- globin (MGB1 or SCGB2A2) was used to specifically detect propanol. Then, it was washed twice in 75% ethanol and resuspended breast cancer cells because it was proven to be a mammary- in RNase-free water. Total RNA was quantified spectrophotometrically specific marker (18). To our knowledge, CEA, CDH1, MUC1/Y, at A260 nm. MUC1/Z, WT1, and CALB2 were never used to detect tumor Reverse transcription-PCR. Reverse transcription was done with the cells in serous effusions by RT-PCR. SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen, Paisley, United Kingdom) using 0.5 Ag total RNA and following the Patients and Methods manufacturer’s instructions. PCR was done in 50 AL reaction mixture containing 1 AL cDNA, Patients. Serous effusions from 114 patients (104 pleural and 10 5 ALof10 PCR buffer (Invitrogen), 1 AL of 10 mmol/L deoxy- ascitic fluids) were prospectively collected from the Departments of nucleotide triphosphate mix (Invitrogen), 1.5 AL of 50 mmol/L Pneumology, Cardiology, and Cancerology of the University Hospital MgCl2, 100 ng of both sense and antisense primers (Table 2), and Center of Saint Etienne and Department of Pneumology of the 1.25 units Taq polymerase (Invitrogen). Primers for h-actin were also University Hospital Center of Lyon, France. All effusions had a reliable used to check RNA integrity and the efficiency of the reverse trans- diagnosis documented by cytologic analysis, histologic analysis of cription step. The sequences of the primers were designed to span pleural biopsy when it was done, and the review of clinical history and splice junctions or in two different exons so that genomic DNA follow-up of the patients without histologic analysis or with conflicting contamination could be monitored. Positive (MCF7 cell line) and results between cytology and histology. Effusions from 43 patients with negative controls (PCR mix without cDNA) were included in each nonmalignant disease caused by infection or cardiac failure served as round of PCR. The primer sequences, their gene localization, and their controls (4 patients had an history of cancer but with a proven benign annealing temperatures are listed in Table 2. After a denaturing step at effusion). Among the 71 malignant effusions, 19 were cytologically 94jC for 3 minutes, PCR was done at 94jC for 1 minute, at the

www.aacrjournals.org 6863 Clin Cancer Res 2005;11(19) October 1, 2005 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis annealing temperature for 1 minute, and at 72jC for 1 minute for combined to MUC1/YZ, we could reach a sensitivity of 81.7% 30 cycles. PCR products were separated by electrophoresis on a 1.3% for malignancy but with a slightly lower specificity (93%). agarose gel. DNA fragments were visualized and photographed under Finally, as expected, MGB1 was significantly more expressed UV light with ethidium bromide staining. in effusions from breast cancer patients than from other Statistical analysis. Marker expression between the different carcinoma (P < 0.0001): 71.4% of breast effusions were populations was compared using the m2 test or the Fisher exact test when theoretical effective was insufficient. The level of statistical positive for MGB1, whereas only one effusion from an ovarian significance was set at P < 0.05. Diagnostic performance of each marker, carcinoma was positive (Table 1). Thus, MGB1 presents a very alone or together, was assessed by computing sensitivity, specificity, good specificity (97.3%) for breast carcinoma effusions. positive predictive value (PPV), and negative predictive value (NPV). Marker expressions in the cytologically negative malignant effusions. Nineteen serous effusions, negative for tumor cell by Results cytologic examination, were diagnosed as malignant by histologic analysis of pleural biopsy and patient clinical data. Evaluation of marker expressions in the whole population of Results of RT-PCR analysis are detailed in Table 4. Briefly, four serous effusions. WT1 and MUC1/REP were highly expressed in samples were negative for all the markers (patients 6, 12, 15, benign effusions (39.5% and 62.8%, respectively). Even if and 16) and three were positive only with MUC1/YZ (patients MUC1/REP was significantly more expressed in malignant than 9, 17, and 18). Considering the malignant-specific markers, 12 in benign serous effusions (P = 0.001), this two markers samples were positive with at least one marker, and two presented a too low specificity to supply a reliable analysis of molecular markers (CEA and Ep-CAM) were necessary and the samples. Moreover, we did not see any correlation between sufficient for the detection of tumor cells in these effusions. The tumor grades and marker expression in breast and pulmonary sensitivity of RT-PCR in cytologically negative malignant effusions (data not shown). effusions was 63.1% (12 of 19) combining CEA and Ep-CAM, Results of marker expressions in the whole population of with a specificity (43 of 43) and a PPV (12 of 12) of 100% effusions are detailed in Table 1. MUC1/Y and MUC1/Z were and a NPV of 86% (43 of 50). Adding MUC1/YZ to this always coexpressed and thus are considered as a unique marker, combination, the sensitivity for malignancy was 78.9% (15 of MUC1/YZ. CALB2 and MUC1/YZ were positive in 2.3% and 19) with 93% specificity (40 of 43), 83.3% PPV (15 of 18), and 7% of benign effusions, respectively. Nevertheless, MUC1/YZ 90.9% NPV (40 of 44). was significantly more expressed in malignant than in benign Moreover, among the three breast carcinoma effusions, two effusions (P < 0.0001). CEA, Ep-CAM, CDH1, and MGB1 were were positive for MGB1, which is of great interest to detect the specifically expressed in malignant effusions. tumor primary site without invasive diagnostic investigations. The sensitivity, specificity, PPV, and NPV of cytology and of Combination of cytology and reverse transcription-PCR. The the molecular markers, alone or in combination, are presented combination of cytology and RT-PCR of CEA + Ep-CAM in the in Table 3. Only the two marker combinations supplying the whole population of effusions allowed to confirm the tumoral higher diagnostic information, in terms of sensitivity and diagnosis in 90.1% of malignant effusions (52 malignant specificity, were included in Table 3. Taken together, CEA, effusions were positive with cytology and 12 more were positive Ep-CAM, CDH1, and MGB1 yielded a positive result in 76% of with RT-PCR) with 100% specificity, 100% PPV, and 86% NPV the malignant effusions with 100% specificity. When they were (Table 3), whereas the combination of cytology and RT-PCR

Ta b l e 2 . Sequences and features of the primers used for RT-PCR

Name Sequence of sense and antisense primers Annealing (°C) Localization Length (bp) MUC1/Y and MUC1/Z 5V-TCTGGTCATGCAAGCTCTAC-3V 56 Exon 2 489 5V-GCACTGACAGACAGCCAAGG-3V Spanning exons 5-6 543 MUC1/REP 5V-ACCACTCTGATACTCCTACC-3V 56 Exon 2 562 5V-GCACTGACAGACAGCCAAGG-3V Spanning exons 5-6 Ep-CAM (TACSTD1) 5V-ACCTGCTCTGAGCGAGTGAGAACCT-3V 64 Exon 3 337 5V-GACCAGGATCCAGATCCAGTTG-3V Exon 7 E-cadherin (CDH1) 5V-GACCATTCAGTACAACGACCCAACC-3V 69 Spanning exons 12-13 605 5V-ACCGCTTCCTTCATAGTCAAACACG-3V Exon16 CEA (CEACAM5) 5V-ACAGTCTCTGCGGAGCTGC-3V 58 Spanning exons 6-7 476 5V-TAGGTCCCGTTATTATTTGGCG-3V Exon 8 Mammaglobin (SCGB2A2) 5V-ACTGCTACGCAGGCTCTGG-3V 56 Spanning exons 1-2 307 5V-TTGCAGCAATCCGTAGTTGG-3V Exon 3 WT1 5V-CTCTTGTACGGTCGGCATCT-3V 56 Exon 7 417 5V-CAGCTGGAGTTTGGTCATG-3V Exon10 Calretinin (CALB2) 5V-CATACTACGGATGTTTGACTTG-3V 56 Spanning exons 6-7 427 5V-TCACGGTCTCTGAGTCTGG-3V Exon11 h-actin 5V-TACCACTGGCATCGTGATGGACT-3V 66 Exon 3 507 5V-TCCTTCTGCATCCTGTCGGCAAT-3V Exon 4

Clin Cancer Res 2005;11(19) October 1, 2005 6864 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Molecular Markers for Cancer Cells in Serous Effusions

Ta b l e 3 . Sensitivity, specificity, PPV, and NPV of cytology and of the molecular markers in the whole population (composed of 71malignant and 43 benign effusions) of effusions

Variables Sensitivity (%) Specificity (%) PPV (%) NPV (%) Cytology 52/71 (73.2) 43/43 (100) 52/52 (100) 43/62 (69.3) CEA 28/71 (39.4) 43/43 (100) 28/28 (100) 43/86 (50) Ep-CAM 47/71 (66.2) 43/43 (100) 47/47 (100) 43/67 (64.2) CDH1 33/71 (46.5) 43/43 (100) 33/33 (100) 43/81 (53.1) MUC1/YZ 33/71 (46.5) 40/43 (93) 33/36 (91.7) 40/78 (51.3) MGB1 11/71 (15.5) 43/43 (100) 11/11 (100) 43/103 (41.7) CEA + Ep-CAM + CDH1 + MGB1 54/71 (76) 43/43 (100) 54/54 (100) 43/60 (71.7) CEA + Ep-CAM + CDH1 + MGB1 + MUC1/YZ 58/71 (81.7) 40/43 (93) 58/61 (95.1) 40/53 (75.5) Cytology + CEA + Ep-CAM 64/71 (90.1) 43/43 (100) 64/64 (100) 43/50 (86) Cytology + CEA + Ep-CAM + MUC1/YZ 67/71 (94.4) 40/43 (93) 67/70 (95.7) 40/44 (90.9)

of CEA + Ep-CAM + MUC1/YZ yielded a sensitivity of 94.4% choice and more elaborated RT-PCR techniques, such as real- with a specificity of 93% for tumor cells detection in malignant time quantitative RT-PCR that are more expensive and effusions. The PPV and NPV were 95.7% and 90.9%, difficult to set. respectively (Table 3). The sensitivities of this two last com- As a first step, we evaluated the expression of the markers in bined test are significantly more elevated than that of cytology the whole population of effusions to determine their interest alone (P < 0.0001 each, Mac Nemar m2 test). for the detection of malignancy. Even if these markers have Markers that could help to distinguish epithelial from non- been studied at the mRNA level, one could expect them to be epithelial malignant effusions. Then, we wanted to know which functional at the protein level according to the numerous molecular markers would undoubtedly help to distinguish publications that have studied them by immunocytochemistry between epithelial and nonepithelial cancer cell in effusions. or flow cytometry in serous effusions (12–14, 16, 19). We Only two markers, CEA and MGB1, were exclusively expressed found two markers highly expressed (WT1 and MUC1/REP) in carcinoma effusions (Table 1). Surprisingly, Ep-CAM was and two other markers rarely expressed (CALB2 and MUC1/YZ) positive in 45% of nonepithelial malignant effusions but was in benign effusions. WT1 and CALB2 are markers of meso- significantly more expressed in carcinoma effusions than in thelial lineage, and as normal desquamated mesothelial cells nonepithelial malignant ones (P = 0.018). can be present in any kind of effusion, they are of limited value, Considering the malignant effusions that were positive for as molecular markers, for the distinction of mesothelioma tumor cells with the combination of cytology and CEA + effusions from other malignant or benign effusions. However, Ep-CAM RT-PCR (64 effusion specimens), 48 had an epithelial CALB2 was significantly more expressed in mesothelioma than origin and 16 a nonepithelial origin. In this population, the in carcinoma effusions. Thus, CALB2 might help to orient a sensitivity, specificity, PPV, and NPV of CEA for carcinoma cells diagnosis in favor of a mesothelioma only when the tumoral are 58.3% (28 of 48), 100% (16 of 16), 100% (28 of 28), and nature of the effusion is ascertained and if it is used in 44.4% (16 of 36), respectively. Combined with MGB1, the combination with confirmed epithelial markers to exclude a sensitivity reached 68.7% (33 of 48) and the NPV was 51.6% maximum of false-positive results. In some immunocyto- (16 of 31). Thus, CEA and MGB1 are very interesting molecular chemical studies, CALB2 was positive in most of benign markers to determine the epithelial origin of tumor cells in effusions (94-100%; refs. 12, 20). It is difficult to compare serous effusions. these results with ours, because the expression of CALB2 was Moreover, CALB2 was significantly more expressed in detected at two different levels (protein or mRNA). The mesothelioma than in carcinoma effusions (P = 0.018). Among mRNA of CALB2 might be an unstable mRNA, but it is also a the four carcinoma effusions that were positive for CALB2, possibility that the polyclonal antibodies used in those studies three were also positive for CEA, thus excluding the possibility produced nonspecific reactions. Moreover, in another immu- of a mesothelioma. Among the five CALB2-positive mesothe- nocytochemical study using a different polyclonal anti-CALB2, lioma, three were suspected to be carcinoma by cytology and only 27% of the benign effusions were stained (21). MUC1/ were diagnosed as mesothelioma by histology. REP, which is the full-length isoform of MUC1 (22), has been described to be expressed in nearly all normal and malignant Discussion epithelial cells (23) but also in normal and malignant hematopoietic cells (24, 25). This is probably why we found We aimed to evaluate an ancillary method that could it expressed in benign serous effusions. Although MUC1/REP improve the yield of cytologic analysis for cancer diagnosis in was significantly more expressed in malignant effusions, which serous effusions. Thus, we focused our study on a rapid, easy, is in agreement with Yu et al. (9), we think that it was too inexpensive, and sensitive technique that could be done in frequently expressed in benign samples to be a reliable most analysis laboratories. Basic RT-PCR strategy seemed to molecular marker for malignancy. Thus far, the expression of be the best compromise between immunocytologic techni- alternatively spliced forms of MUC1 has been poorly studied. ques that often lacks sensitivity and flexibility in the markers MUC1/Y and MUC1/Z have been detected in primary breast

www.aacrjournals.org 6865 Clin Cancer Res 2005;11(19) October 1, 2005 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis tumors by Western blot analysis (15, 26) and in ovarian cancer Moreover, Ep-CAM was expressed in most carcinomas (74.5%) tissue by RT-PCR (27). Moreover, MUC1/Y has been detected and mesothelioma effusions (57.1%). Indeed, some reports by flow cytometry on the surface of ovary and breast malignant described infrequent and weak expression of Ep-CAM in cells in effusions (19). We showed here that MUC1/Y and mesothelioma by immunocytochemistry (13, 29, 32). Thus, MUC1/Z were mostly expressed in malignant effusions of although it was significantly more expressed in carcinoma than epithelial and nonepithelial origin and poorly expressed in in nonepithelial malignant effusions, we rather consider it as a benign effusions. They presented a very good specificity for tumor than a carcinoma marker in serous effusions. Therefore, malignancy (93%), and they were expressed alone in some if Ep-CAM seems to be of limited interest to distinguish malignant effusions, contributing to enhance the sensitivity of carcinoma from mesothelioma effusions, it is a valuable the RT-PCR combined test. Thus, they could be interesting molecular marker for the detection of tumor cells in serous molecular markers to add in some panels of markers to effusions. CDH1, an epithelial-specific adhesion molecule, has enhance the detection sensitivity of tumor cells, keeping in been proven to reliably distinguish carcinoma cells from benign mind that they are not totally discriminant. reactive mesothelial cells, but not from mesothelioma cells, in The other markers (CEA, Ep-CAM, CDH1, and MGB1) were serous effusions by immunocytology (12, 20, 33). We also specifically expressed in malignant effusions, therefore being found that CDH1 was specifically expressed in malignant very good candidates for molecular detection of malignancy in epithelial and nonepithelial serous effusions, but it was not serous effusions. CEA was often used in immunohistochemical very sensitive. It was expressed alone in none of the cyto- studies to distinguish carcinoma cells from reactive mesothelial logically negative malignant effusions and in only one cyto- or mesothelioma cells (14, 28). However, it was never used as a logically positive malignant effusion. That is why its usefulness RT-PCR marker in serous effusions. In this study, we showed in a panel of molecular markers is relative. MGB1 was reported that CEA was not only specifically expressed in malignant previously as a specific marker of breast carcinoma cells (18). effusions but was also restricted to the epithelial malignancies. However, using nested RT-PCR, Gru¨newald et al. (34) recently This makes it an exceptional molecular marker to distinguish found MGB1 expression in endometrial and ovarian carcinoma epithelial from nonepithelial malignant serous effusions. effusions as well as in stomach and lung effusions in another Ep-CAM is generally admitted as an epithelial marker expressed study (11). In this last study, they also found MGB1 expression on the surface of most carcinomas and it was often used by in 12.9% of their control effusions. We only found MGB1 immunocytochemistry to distinguish carcinoma from meso- expression in breast and in one effusion from an ovarian thelioma or reactive mesothelial cells in serous effusions carcinoma, and all our control effusions were negative. We (13, 29, 30). Using real-time RT-PCR, Nagel et al. found it think that the highest sensitivity of nested RT-PCR could weakly expressed in cultured mesothelial cells isolated from explain the differences between our respective studies. With ascites of patients with benign disease (10). In our study, it was 97.3% of specificity for breast carcinoma effusions, MGB1 is not expressed in benign serous effusions. This could be because therefore a marker of choice in the search of primary site in conventional RT-PCR is less sensitive than real-time RT-PCR carcinoma effusions of unknown origin. but also because their mesothelial cells were cultivated with Then, after the determination of the most efficient markers for FCS, which contained numerous cytokines susceptible to the detection of tumor cells in serous effusions, we evaluated stimulate Ep-CAM expression as tumor necrosis factor-a (31). their interest as an adjunctive RT-PCR test to cytology. Therefore,

Ta b l e 4 . Marker expressions in cytologically negative malignant effusions

Patients Age/sex Diagnosis CEA Ep-CAM CDH1 MUC1/YZ MGB1

168/FBreastcarcinoma + 264/FBreastcarcinoma ++ + + 376/FBreastcarcinoma +++ + 4 66/M Lung adenocarcinoma + + + + 5 70/F Lung adenocarcinoma + + + + 6 63/M Small cell lung carcinoma 7 71/M Small cell lung carcinoma + 8 62/M Nasopharyngeal carcinoma ++ + 9 74/M Nasopharyngeal carcinoma + 10 56/M Rectum carcinoma + + 11 64/F Stomach carcinoma + + + + 12 79/M Stomach carcinoma 13 55/M Pancreatic carcinoma + 14 73/F Unknown carcinoma + + + 15 54/F Sarcoma 16 79/M Sarcoma 17 65/M Lymphoma + 18 48/M Lymphoma + 19 75/M Mesothelioma +

Clin Cancer Res 2005;11(19) October 1, 2005 6866 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Molecular Markers for Cancer Cells in Serous Effusions

we studied their expression in a subgroup of cytologically abrogate most of markers’ ectopic expression staying enough negative malignant effusions. We found that there were two sensitive to be clinically useful to complement cytologic pertinent marker combinations: CEA + Ep-CAM + MUC1/YZ examination. Adding RT-PCR to clinical practice for the combination yielding the highest sensitivity (78.9%) with a very diagnosis of serous effusions would represent a nonexcessive good specificity (93%) and CEA + Ep-CAM combination extra workload in terms of time, cost, and manpower compared yielding the highest specificity (100%) with a good sensitivity with the benefits that it would bring to the patients. (63.1%). The two marker combinations are interesting. How- Finally, we also showed that CEA and MGB1 were specifically ever, our standpoint is that specificity is more important than expressed in epithelial malignancies and that, among the 64 sensitivity for a diagnosis of malignancy that has to be sure. effusions detected as malignant with the combined test of Moreover, such molecular biology techniques have to be cytology and CEA + Ep-CAM RT-PCR, they allowed to reliable to avoid invasive diagnostic techniques. Thus, the determine the epithelial origin of 68.7% of carcinoma effusions combination of CEA and Ep-CAM as an ancillary RT-PCR test with 100% specificity. in cytologically negative and suspicious serous effusions seems In conclusion, this study showed that CEA and MGB1 are to us the best choice. Moreover, we would also complete this test very reliable molecular markers for the detection of carcinoma with MGB1 because it can predict a breast or, less frequently, an cells in serous effusions and that a RT-PCR test of CEA, ovarian origin of the effusion. Ep-CAM, and MGB1 could be a helpful adjunct to cytologic Then, we showed that the combination of cytology and examination for the diagnosis of malignancy and the orienta- CEA + Ep-CAM RT-PCR test improved the diagnosis accuracy, tion of the tumor origin. yielding 90.1% sensitivity, 100% specificity, 100% PPV, and 86% NPV for malignancy. These results are much better than the 81.1% sensitivity and 70.1% specificity obtained by Fiegl Acknowledgments

et al. (11) combining cytology, fluorescence in situ hybridiza- We thank C. Ruesch, M.D., P. Sagnol, M.D., and J.M.Vergnon, M.D., Ph.D., for tion for aneuploidy, and nested RT-PCR of MGB1 and MGB1-B. providing pleural effusions; S. Laporte and E. Preysle from DRCC of St. Etienne for Moreover, it seems that basic RT-PCR strategy allowed to statistical analysis, and D. Bernard-Gallon, Ph.D., for critical reading of the article.

References 1. Monte SA, Ehya H, Lang WR. Positive effusion cyto- 12. Kitasume H, Kitamura K, Mukai K, et al. Cytologic 24. Brugger W, Buhring HJ, Grunebach F, et al. Expres- logy as the initial presentation of malignancy. Acta differential diagnosis among reactive mesothelial cells, sion of MUC-1 epitopes on normal bone marrow: Cytol 1987;31:448 ^ 52. malignant mesothelioma, and adenocarcinoma. implications for the detection of micrometastatic 2. Bedrossian CWM. Diagnostic problems in serous Cancer 2000;90:55 ^ 60. tumor cells. J Clin Oncol 1999;17:1535 ^ 44. effusions. Diagn Cytopathol 1998;19:131^ 7. 13. Kuhlmann L, Berghauser KH, Schaffer R. Distinction 25. Treon SP, Mollick JA, Urashima M, et al. Muc-1 3. Fetsch PA, Abati A. Immunocytochemistry in effu- of mesothelioma from carcinoma in pleural effusions. core protein is expressed on multiple myeloma cells sion cytology: a contemporary review. Cancer 2001; Animmunocytochemical studyon routinelyprocessed and is induced by dexamethasone. Blood 1999;93: 93:293^ 308. cytoblock preparations. Pathol Res Pract 1991;187: 12 87 ^ 9 8 . 4. Mohanty SK, Dey P. Serous effusions: diagnosis of 467^71. 26. Zrihan-Licht S,Vos HL, Baruch A, et al. Character- malignancy beyond cytomorphology. An analytic 14. Ghosh AK, Gatter KC, Dunnill MS, Mason DY.Immu- ization and molecular cloning of a novel MUC1 protein, review. Postgrad Med J 2003;79:569 ^ 74. nohistological staining of reactive mesothelium, meso- devoid of tandem repeats, expressed in human breast 5. Gerhard M, Juhl H, Kalthoff H, Schreiber HW, thelioma, and lung carcinoma with a panel of cancer tissue. EurJ Biochem 1994;224:787 ^ 95. Wagener C, Neumaier M. Specific detection of carci- monoclonal antibodies. J Clin Pathol 1987;40:19^25. 27. ObermairA, Schmid BC, Stimpfl M, et al. Expression noembryonic antigen-expressing tumor cells in bone 15. Baruch A, Hartmann M, Zrihan-Licht S, et al. Prefer- of MUC1splice variants in benign and malignant ovar- marrow aspirates by polymerase chain reaction. J Clin entialexpressionof novel MUC1tumorantigenisoforms ian tumours. Int J Cancer 2002;100:166 ^ 71. Oncol1994;12:725^9. inhumanepithelialtumors andtheir tumor-potentiating 28. Lee JS, Nam JH, Lee MC, Park CS, Juhng SW. 6. Mori M, Mimori K, Ueo H, et al. Clinical significance function. IntJCancer1997;71:741 ^ 9. Immunohistochemical panel for distinguishing bet- of molecular detection of carcinoma cells in lymph 16. HechtJL, Lee BH, PinkusJL, Pinkus GS.The value of ween carcinoma and reactive mesothelial cells in nodes and peripheral blood by reverse transcription- Wilms tumor susceptibility gene1in cytologic prepara- serous effusions. Acta Cytol 1996;40:631 ^ 6. polymerase chain reaction in patients with gastroin- tions as a marker for malignant mesothelioma. Cancer 29. Sheibani K, Shin SS, Kezirian J, Weiss LM. Ber-EP4 testinal or breast carcinomas. J Clin Oncol 1998;16: 2002;96:105 ^ 9. antibody as a discriminant in the differential diagnosis 128^ 32. 17. Doglioni C,Tos AP, Laurino L, et al. Calretinin: a novel of malignant mesothelioma versus adenocarcinoma. 7. Okamoto I, MorisakiT,SasakiJ, et al. Molecular detec- immunocytochemical marker for mesothelioma. Am J Am J Surg Pathol 1991;15:779 ^ 84. tionofcancercellsbycompetitivereversetranscription- Surg Pathol1996;20:1037^46. 30. Latza U, Niedobitek G, Schwarting R, Nekarda H, polymerase chain reaction analysis of specific CD44 18. FlemingTP,Watson MA. Mammaglobin, a breast- Stein H. Ber-EP4: new monoclonal antibody which variant RNAs. JNatl Cancer Inst1998;90:307 ^15. specific gene, and its utility as a marker for breast can- distinguishes epithelia from mesothelia. J Clin Pathol 8. Yang CT, Lee MH, Lan RS, Chen JK. Telomerase cer. Ann N YAcad Sci 2000;923:78 ^ 89. 19 9 0;4 3 :213 ^ 9. activity in pleural effusions: diagnostic significance. 19. HartmanM,BaruchA,RonI,etal.MUC1isoform 31. Grunberg E, Eckert K, Karsten U, Maurer HR. Effects JClin Oncol1998;16:567^73. specific monoclonal antibody 6E6/2 detects prefer- of differentiation inducers on cell phenotypes of cul- 9. Yu CJ, Shew JY, Liaw YS, et al. Application of mucin ential expression of the novel MUC1/Y protein in tured nontransformed and immortalized mammary quantitative competitive reverse transcription poly- breast and ovarian cancer. Int J Cancer 1999;82: epithelial cells: a comparative immunocytochemical merase chain reaction in assisting the diagnosis of 256 ^ 67. analysis.Tumour Biol 2000;21:211^ 23. malignant pleural effusion. Am J Respir Crit Care Med 20. He DN, Zhu HS, Zhang KH, et al. E-cadherin and 32. Ordonez NG.Value of the Ber-EP4 antibody in differ- 2001;164:1312^ 8. calretinin as immunocytochemical markers to differen- entiating epithelial pleural mesothelioma from adeno- 10. Nagel H, Werner C, Hemmerlein B. Real-time tiate malignant from benign serous effusions.World J carcinoma. The M.D. Anderson experience and a reverse transcription-polymerase chain reaction assay Gastroenterol 2004;10:2406 ^ 8. critical review of the literature. Am J Clin Pathol 1998; for GA733-2 m RNA in the detection of metastatic 21. Simsir A, Fetsch P, Mehta D, Zakowski M, Abati A. 109 :85 ^ 9. carcinoma cells in serous effusions. Am J Clin Pathol E-cadherin, N-cadherin, and calretinin in pleural effu- 33. Schofield K, D’AquilaT, Rimm DL.The cell adhesion 2003;120:888 ^ 901. sions: the good, the bad, the worthless. Diagn Cyto- molecule, E-cadherin, distinguishes mesothelial cells 11. Fiegl M, Haun M, MassonerA, et al. Combination of pathol 1999;20:125^ 30. from carcinoma cells in fluids. Cancer1997;81:293^ 8. cytology, fluorescence in situ hybridization for aneu- 22. Gendler SJ. MUC1, the renaissance molecule. 34. Gru«newald K, Haun M, Fiegl M, et al. Mammaglobin ploidy, and reverse-transcriptase polymerase chain JMammary Gland Biol Neoplasia 2001;6:339 ^ 53. expression in gynecologic malignancies and malignant reaction for human mammaglobin/mammaglobin B 23. Taylor-Papadimitriou J, Burchell J, Miles DW, Dalziel effusions detected by nested reverse transcriptase- expression improves diagnosis of malignant effusions. M. MUC1 and cancer. Biochim Biophys Acta 1999; polymerase chain reaction. Lab Invest 2002;82: J Clin Oncol 2004;22:474 ^ 83. 14 55 :3 01 ^ 13. 1147 ^ 53.

www.aacrjournals.org 6867 Clin Cancer Res 2005;11(19) October 1, 2005 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2005 American Association for Cancer Research. Evaluation of a Panel of Molecular Markers for the Diagnosis of Malignant Serous Effusions

Karine Passebosc-Faure, Guorong Li, Claude Lambert, et al.

Clin Cancer Res 2005;11:6862-6867.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/19/6862

Cited articles This article cites 34 articles, 9 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/19/6862.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/19/6862.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/19/6862. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.