INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: malignant pleural effusion; mucins; reverse transcriptase polymerase chain reaction

Pleural effusions are important and commonly occurring complications produced by a wide variety of diseases. Approximately 20% of pleural effusions are caused by malignancy, and, in 10 to 50% of cancer patients, may be the initial presentation (1). Establishing the diagnosis of a malignant pleural effusion indicates advanced disease and is associated with limited survival (2). It is then clinically important to clarify the precise cause of the pleural effusion, especially to discriminate benign from malignant effusions. Thoracentesis with cytologic and biochemical examinations is the initial diagnostic approach to pleural effusion.

Although cytologic examination is the most informative laboratory procedure for diagnosis of malignant pleural effusions, it may fail to identify 20 to 60% of malignant pleural involvement (3, 4). Factors influencing the diagnostic yield of pleural effusion cytology include the sample volume, number of malignant cells in the pleural fluid, and the experience of the cytopathologist. Usually, more than one thoracentesis are required to prove that the effusion is malignant (5). Further invasive techniques, such as pleural biopsy, thoracoscopy, or thoracotomy may be required to achieve the diagnosis. Adjunct methods, such as cytogenetics, biochemical parameters, and tumor markers in the pleural fluid have been intensively evaluated to distinguish between malignant and benign effusions (3, 4, 6, 7). Yet, despite these diagnostic approaches, definitive diagnosis still could not be confirmed in approximately 20% of patients with malignant pleural effusions (3).

In search of a better method to diagnose malignant pleural effusions, the role of mucins is investigated. Mucins are highly glycosylated, high-molecular-weight (> 200 kD) proteins present on the surface of many epithelial cells. Alteration in expression and post-translational modification of mucins is frequently shown in tumors, and may contribute to the process of cancer invasion and metastasis (8). So far, 10 distinct gene loci of mucins have been identified in human beings, and are named MUC1, -2, -3, -4, -5B, -5AC, -6, -7, -8, and -9 (9). Studies comparing the expression of mucin genes in normal and malignant tissues have demonstrated organ specificity of mucin gene expression in normal tissues and the extensive loss of gene regulation after malignant transformation (17). Abnormal increases in the concentration of mucin and mucin- associated antigens in malignant pleural effusions have been demonstrated, probably owing to the presence of aberrantly transformed epithelial cells, especially adenocarcinoma cell type (22, 23).

Recently, the use of reverse transcriptase/polymerase chain reaction (RT-PCR) to identify cell-specific messenger RNA, including that of mucin genes, has shown to be a highly sensitive tool in the detection of minimal involvement of blood, bone marrow, and lymph node in a variety of malignancies (24). Previously, we had developed a quantitative competitive (QC) RT-PCR method to quantify the expression of mucin messenger RNA (mRNA) (MUC1, MUC2, and MUC5AC) (27, 28). In this study, we evaluated the efficiency of mucin QC RT-PCR in the detection of exfoliated cancer cells in pleural effusions and demonstrated these tests as potentially useful in assisting the diagnosis of malignant pleural effusion.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A more detailed account of the methods used in this study is available in the online data supplement.

Patients and Specimens

From June 1997 to May 1998, exudative pleural fluids were obtained prospectively from 112 consecutive patients who were referred to our ultrasonography department for diagnostic examination before treatment. Fluids with frank pus were excluded. All pleural effusions had definite etiologies documented by examination of effusion biochemistry, cytology, pleural biopsy, percutaneous biopsy, endoscopic examination with biopsy, and clinical follow-up. The underlying diseases of these patients are shown in Table 1. Fifty-four were malignant effusions confirmed by cytologic examination. Thirty-five were from patients with benign pulmonary diseases diagnosed by laboratory (bacterial, cytologic, and histologic diagnosis) and clinical findings. Twenty-three were from patients with pulmonary or extrapulmonary malignancies, but lacked cytologic evidence of malignant cells in pleural fluid after at least two examinations. Eleven of these effusions were later confirmed to be malignant (including five lung adenocarcinomas, one lung squamous cell carcinoma, one small cell lung carcinoma, one hepatocellular carcinoma, one renal cell carcinoma, one hypopharyngeal carcinoma, and one prostate adenocarcinoma; three of these were diagnosed by pleural biopsy, three by thoracoscopy, and five by clinical follow-up). Twelve were determined to be caused by benign process (five caused by bacterial pneumonia, three hepatic hydrothorax, one pyogenic liver abscess, one acute pancreatitis). For these, the underlying malignancies were determined to be: four hepatocellular carcinomas, two lung adenocarcinomas, one lung squamous cell carcinoma, one small cell lung carcinoma, two laryngeal carcinomas, one renal cell carcinoma, one pancreatic adenocarcinoma.

The clinical data of each patient were collected by chart review of hospitalization and outpatient medical records. All samples were obtained during the course of routine diagnostic evaluation. Cytologic analysis was performed following normal clinical procedures.

RNA Extraction

After the addition of heparin, 20 to 40 ml of pleural fluids were centrifuged at 400 × g for 10 min at room temperature, and the supernatants were discarded. Cells were collected, washed with 1 ml phosphate-buffered saline (pH 7.4) at 4° C, and subjected to RNA extraction. The RNA was washed with 75% alcohol, resuspended, and quantified spectrophotometrically.

QC RT-PCR

The competitive templates for amplification of specific mucin genes were complementary RNA (cRNA) fragments produced by in vitro transcription of mutated DNA fragments constructed as described previously (28).

Six micrograms of total RNA from pleural effusion and 8 µg of random hexamer were mixed and aliquoted into eight tubes, each containing serially diluted target gene RNA internal standards. These were then subjected to RT-PCR. The PCR products were electrophoresed in nondenaturing polyacrylamide gel, and stained in ethidium bromide solution. The signal intensity of amplified native and mutated products was measured and digitized by densitometer. The expression ratios of mucin genes were defined as the calculated amounts of mucin gene RNA transcripts divided by the amounts of G-like gene transcripts (28).

Statistical Analysis

Statistical evaluation was performed by computer analysis with SPSS Software (SPSS Inc., Chicago, IL). The Mann-Whitney U test was used to estimate the significance of the differences in mucin gene expression ratios between malignant and benign samples. The level of statistical significance was set at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The first part of this study consisted of 54 cases of cytologically positive malignant pleural effusion and 35 cases of benign exudative pleural effusion in order to determine the cut-off values of mucin QC RT-PCR. Examples of QC RT-PCR for MUC1, MUC2, MUC5AC, and G-like genes in a case with malignant pleural effusion resulting from lung adenocarcinoma are shown in Figure 1.

View larger version (27K): [in this window] [in a new window]   Figure 1.   Examples of the QC RT-PCR products of MUC1, MUC2, MUC5AC, and G-like genes after electrophoresed in 6% nondenaturing polyacrylamide gel and stained with ethidium bromide. This is the specimen obtained from a patient with lung adenocarcinoma.

The mean, standard deviation, median, and ranges of expression ratios of three mucin genes in patients with benign and malignant pleural effusions are presented in Table 2. The expression amounts of MUC1 mRNA are significantly higher in pleural fluids with detectable malignant cells than in fluids of benign diseases (p < 0.000). Similar results are also shown in evaluation of MUC5AC mRNA expression ratio (p < 0.000), but not in that of MUC2 (p = 0.124). (Figure 2).

View larger version (14K): [in this window] [in a new window]   Figure 2.   Scattergram of the distribution of the expression ratios of MUC1 (A), MUC2 (B), and MUC5AC (C ) in pleural effusion of different groups. Open circles: effusion samples with cytology positive for malignancy (M, n = 54); open triangles: effusion samples with benign diseases (B, n = 35). The horizontal lines indicate the cutoff values for differentiating malignant from benign effusions.

The cutoff values were defined as the benign group's mean + 2 SD (95% percentile), and were 0.126 for MUC1, 0.028 for MUC5AC. The test sensitivity and specificity of MUC1 expression ratio were 63.0% (34/54) and 94.3% (33/35), respectively; for MUC5AC expression ratio, these values were 72.2% (39/ 54) and 94.3% (33/35), respectively.

MUC1 ratios above cutoff value were found in 29 of 43 adenocarcinomas (21 of 34 lung cancers, four of four gastric cancers, three of four breast cancers, and one ovarian cancer), one of four squamous cell carcinomas (one tongue cancer, none of the two nasopharyngeal cancers, and one buccal cancer), two of four lymphomas, one myeloma, and one melanoma. MUC5AC ratios above cutoff value were detected in 35 of 43 adenocarcinomas (27 of 34 lung cancers, in all of gastric cancers, breast cancers, and ovarian cancers), one small cell lung carcinoma, two of four squamous cell carcinomas (both were nasopharyngeal carcinomas), and one melanoma.

For benign disease, MUC1 was beyond the threshold in only one of 19 tuberculous pleurisy, and in one of 10 bacterial pneumonia. For MUC5AC, false positives were noted in one case each of tuberculous pleurisy and bacterial pneumonia. With the combination of both tests, sensitivity (defined as the presence of at least one mucin mRNA showing a positive value) could be improved to detect 47 of 54 (87.0%) malignant pleural effusions. The test specificity was 88.6%; positive predictive value (PPV) was 92.2% and negative predictive value (NPV) was 81.6%.

The second part of the study included 23 cases with underlying pulmonary or extrapulmonary malignancies, but with cytologically negative pleural effusion. Eleven of these were later proven to be malignant; the others were considered to be benign. For cytologically negative malignant pleural effusions, the expression ratios of MUC1 and MUC5AC were higher compared with those of benign processes (Table 3, p = 0.002 for MUC1, p = 0.004 for MUC5AC) (Figure 3). The PPV and NPV by applying calculated cutoff values from the first part of study for MUC1 were 100% (7/7), and 75% (12/16), respectively. The calculated PPV and NPV for MUC5AC were the same: 100% (7/7) and 75% (12/16). By combining MUC1 and MUC5AC (MUC1 + MUC5AC) together, the PPV was 100% (9/9), and the NPV was 85.7% (12/14).

View larger version (11K): [in this window] [in a new window]   Figure 3.   Scattergram of the distribution of the expression ratios of MUC1 and MUC5AC in patients with malignancies whose effusions were cytologically negative. Open circles: effusion samples later proved to be related to malignant process (M, n = 11); open triangles: effusion with benign processes (B, n = 12). The horizontal lines indicate the cutoff values for differentiating malignant from benign effusions.

The diagnostic sensitivity of mucin gene QC RT-PCR tests on the whole population of malignant effusions (54 cytologically positive effusions and 11 cytologically negative effusions), as well as in various groups of patients classified by histologic origin of their cancer is summarized in Table 4. Patients with nonepithelial tumors tended to have lower probability for MUC5AC expression than that of carcinomas. In comparing the two major cell types of carcinomas, the sensitivity of MUC5AC test was higher for adenocarcinomas than for squamous cell carcinomas.

By comparing the age, sex distribution, and laboratory tests (cell counts, subpopulation of differential cells, contents of total protein, and lactate dehydrogenase) of pleural fluids among patients with positive and negative results, no parameter could predict the test positivity of mucin QC RT-PCR.

The diagnostic sensitivity, specificity, PPV, and NPV of the tests for the whole population of 112 patients are listed in Table 5. In combined evaluation with both MUC1 and MUC5AC tests, the sensitivity, specificity, PPV, and NPV were 86.1%, 91.5%, 93.3%, and 82.7%, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study demonstrates the usefulness of mucin QC RT-PCR examination in assisting the diagnosis of malignant pleural effusions. Many biochemical markers had been evaluated as an adjunctive method in assisting the diagnosis of malignant pleural effusions. These methods included carcinoembryonic antigen (CEA), cytokeratin fragment 19 (CYFRA-21-1), squamous cell carcinoma antigen (SCC), carbohydrate antigen 19-9 (CA 19-9), carbohydrate antigen 15-3 (CA 15-3), cancer antigen 125 (CA 125), sialic acid, sialyl stage-specific embryonic antigen (SSEA-1), neuron-specific enolase, phosphohexose isomerase, and 2-microglobulin (23, 29). Higher concentrations of these markers have been observed in malignant effusions than in fluids of benign processes. The test sensitivity and specificity ranged around 31 to 95% and 61 to 95%, respectively, with accuracy at 65 to 85%. However, because there was significant overlap between malignant and benign effusions, these tests were generally not considered useful diagnostically.

Recent studies demonstrated that PCR method targeted at tumor-specific genetic abnormalities could detect a small number of cancer cells mixed within a large number of normal cells (24). By PCR amplification, only a scanty amount of RNA (much less than those needed for Northern blot analysis and slot blot analysis) was required to analyze the expression of a specific gene. This method offered considerable promise for early diagnosis of cancer and detection of tumor progression before clinically evident metastasis. However, the clinical application has been limited so far. The primary limitation had been the lack of suitable target genes common to most solid tumors. By analyzing specific CD44 variant RNAs, Okamoto and coworkers reached a sensitivity of 75.4%, and a specificity of 96.2% in detecting lung cancer and bladder cancer cells in pleural effusion and urine (32). Surfactant protein A (SP-A) gene had also been used as a target for analysis. Its gene transcript could be detected merely in patients with lung adenocarcinoma, while not in other types of lung cancer, nor in any benign diseases, such as pneumonia (33). Because mucin genes expressed mainly in epithelial cells and frequently overexpressed in cancer cells, it would be reasonable to anticipate its applicability in assisting the differential diagnosis of pleural fluids. In this study, the sensitivity and specificity of MUC1, MUC5AC, and MUC1 + MUC5AC QC RT-PCR tests were at least as good as or even better than RT-PCR analysis of CD44 and SP-A genes (Table 5).

MUC1 membrane mucin is the most widely expressed mucin in normal epithelial cells and its expression increases drastically when the cells become malignant (34). This has been well-documented in breast and ovarian cancer, as well as in colon, lung, prostate, gastric, and pancreatic cancers. MUC1 mucin has also been shown to be associated with tumor progression and poor clinical outcome (9, 35). Because MUC1 mucin is thought to be restricted to epithelial cells, it has been used for the detection of micrometastatic tumor cells in patients with solid tumors. In this study, we demonstrate the expression of MUC1 not only in pleurisy of carcinomas, but also in those of nonepithelial tumors (melanoma, myeloma, and lymphoma).

Previous studies have shown the expression of MUC1 mRNA and protein in non-Hodgkin's lymphoma, myeloma, melanoma, astrocytoma, neuroblastoma, and even in normal hematopoietic cells (36). The mechanisms responsible for MUC1 mucin overexpression in nonepithelial cells include gene amplification, chromosomal translocation, hypomethylation, and activation by transcription factors. As in epithelial tumors, the aberrant expression of MUC1 in nonepithelial tumors may have a potential role in tumorigenesis and cellular function.

Unlike MUC1, MUC2 and MUC5AC are secretory mucins expressed exclusively in epithelial cells of specific organ tissues. MUC2 is mainly expressed in glandular cells of small intestine, and its overexpression has been shown in cancers of lung, colon, stomach, and pancreas (19, 39). MUC5AC is expressed mainly in airway (both nasal and tracheal epithelium) and stomach, with overexpression having been demonstrated in cancers of lung, stomach, breast, and pancreas (18- 21, 27, 40). The expression of these two genes in pleural fluids may indicate the existence of epithelial cells with specific origins.

In this study, the expression ratio of MUC2 was similar in both malignant and benign groups. However, the expression of MUC5AC was significantly higher in the malignant group, mainly in the adenocarcinoma cell type. As pleural effusions caused by carcinomas of the lung, nasopharynx, and stomach represented the majority of the studied population, the result reflected the cell origin of the malignancies. However, MUC5AC expression could also be detected in melanoma (one case) and renal cell carcinoma (one case). So far, the information regarding mucin gene expression in these two cell types is scarce. Further studies would be necessary to determine whether the expression of secretory mucin in melanoma or renal cell carcinoma plays any role in tumorigenesis, or is just an epiphenomenon.

Besides MUC1, 2, and 5AC genes, the other seven mucin genes (MUC3, 4, 5B, 6, 7, 8, 9) are also good candidates for investigation. Because MUC4 and MUC8 genes recently have been shown to be constantly present in airway epithelium and in lung carcinoma (40, 41), further studies should be carried out to evaluate the applicability of these two particular genes in serving as diagnostic adjuncts for pleural effusion.

In this study, we used QC RT-PCR instead of conventional RT-PCR to quantify the amounts of gene transcripts. The limitation of RT-PCR is related to laboratory method per se. Tube-to-tube variation in reaction efficiency may cause as much as a sixfold difference among reactions in both reverse transcription and polymerase chain reaction, making precise quantification impossible in RT-PCR.

Several modifications have been made to overcome this problem. These include kinetic PCR assay (using serial dilution of RNA templates and cycle number restriction to control the PCR reaction at the exponential phase) and quantitative PCR (adding external or internal standards of target gene for competition throughout the reaction of RT-PCR). Among these modifications, constructing a mutant RNA fragment of the target gene (internal standard) to compete with the native target gene transcripts during RT-PCR has been found to be the best choice for eliminating many factors that interfere with the interpretation of data. This method overcomes uncertain factors in many steps that may affect efficiency, including RNA extraction, complementary DNA (cDNA) synthesis, and PCR amplification. It also has a low test variability, usually less than 15%. Previously, we had constructed cRNA internal standards from non-tandem repeat coding sequences of MUC1, MUC2, and MUC5AC, and showed an acceptable variability in quantification. Because of the low data variability, the low number of templates needed, and no need of radioactive material, QC RT-PCR would be an ideal method to evaluate gene expression in clinical specimens.

However, there are several disadvantages to the QC RT-PCR method. First, reactions must be carried out in multiple tubes (eight tubes in the present study) containing serial dilution of competitive fragments, a design developed to establish a competition curve and to assure the quality of reaction. Both the cost and the time consumed in examining gene expression are high, because at least two sets of RT-PCR must be performed for the target gene in each case: one set for the gene of interest and the other for the internal control gene. Second, the uncertainty of potential differences in the kinetics of amplification between mutant and native fragments is also a problem frequently encountered.

Recently, real-time quantitative RT-PCR (RTQ RT-PCR) to kinetically quantify the PCR products when the amplification reaction is still in the exponential phase has been developed. RTQ RT-PCR is shown to be a more reliable and sensitive method than QC RT-PCR in quantifying the starting copy number of target molecules when compared with the end product method used in the present study. However, a comparison between these two methods shows almost comparable accuracy in detecting relative expression of target mRNA among samples (42). Furthermore, high unit cost (much higher than QC RT-PCR) also hinders the wide applicability of RTQ RT-PCR for clinical purposes.

This study also compared the usefulness of mucin QC RT-PCR with cytologic analysis. Frequently, MUC1 and MUC5AC mRNAs were expressed in the pleural fluids of patients exhibiting malignant cytology. However, the more striking fact was the demonstration of detectable mucin mRNA in nine of 11 cytologically negative malignant pleural effusions. In fact, the use of cytology or mucin PCR, or both, confirmed the diagnosis of 56 of 65 malignant effusions (86.1%). The presence of only a small number of tumor cells, at a level too low for routine cytologic diagnosis, may explain the fact that some pleural effusions demonstrated mucin mRNAs while found to be cytologically negative.

Thus, PCR examination of specific target genes provides a noninvasive assay sensitive enough to alert and help cytopathologists when cytologic examination remains suspect or negative. It also constitutes a good argument for physicians to rapidly perform another thoracentesis or thoracoscopy.

False positives of mucin QC RT-PCR were encountered in four cases, two of tuberculous pleurisy and two of parapneumonic effusion. Previously, it has been shown that infection or inflammation in the pleural space could render falsely increased tumor markers (43, 44). The mechanism causing the false elevation of markers is unknown, but has been speculated to be the alteration of protein markers by bacterial enzymes. However, the aforementioned mechanism cannot explain the falsely increased mucin mRNA in benign inflammatory fluids. Mucin genes are expressed in low amounts in nonepithelial cells, including inflammatory cells (37, 45). The transcription of mucin genes may be further activated by the inflammatory process (13). Therefore, the elevated mRNA of mucin genes of benign exudates in our study may be due to the expression of mucins by nonepithelial cells in the pleural fluids.

In conclusion, mucin QC RT-PCR analysis is considered very useful in assisting the diagnosis of malignant pleural effusion, not only in lung cancer, but also in other types of cancerous pleurisy.