Context.—We previously found telomere repeat amplification protocol (TRAP) in situ helpful in the diagnosis of malignancy in effusions, whereas varying sensitivities and specificities for malignancy were reported by investigators using extract-based TRAP.
Objective.—To compare the 2 methods and to elucidate the discrepancies between them.
Design.—Twenty-three effusions were analyzed. Telomerase activity of whole cell lysate was measured with a Telo TAGGG telomerase polymerase chain reaction ELISA PLUS kit with modifications to exclude polymerase chain reaction inhibitors. TRAP in situ was performed on cytospins. An estimate of total TRAP activity in the specimen was made based on the amount of positive cells, their fluorescence intensity, and the proportion of different cell types in the specimen. The estimate was compared with the level of telomerase activity in cell lysate–based TRAP.
Results.—TRAP in situ: Thirteen of 14 malignant cases and 2 of 2 equivocal cases showed moderate/strong reactivity. Five of 7 benign effusions were negative; in 2 of 7, mesothelial cells showed weak reactivity. Cell lysate–based TRAP assay: In 4 cases no internal standard was detected, indicating the presence of polymerase chain reaction inhibitors. The relative telomerase activities were 33.1 to 72.7 with a considerable overlap between malignant (48 ± 9, mean ± SD) and benign (43 ± 9) cases.
Conclusions.—The TRAP in situ results correlated to final diagnoses, whereas the cell lysate–based TRAP assay did not differentiate between malignant and benign cases. The varying proportions of positive cells and the variation in fluorescence intensity in the TRAP in situ slides explained some of the discrepancies. The problems encountered with TRAP performed on cell lysates are partly overcome using TRAP in situ.
The diagnosis of effusions is a challenge to the cytopathologist. Morphologic criteria do not always allow the distinction between malignant cells and benign mesothelial cells and the complexity of the cellular content often makes it difficult to identify a small number of malignant cells. Therefore, a general marker of malignancy would provide valuable information.
Telomerase is a ribonucleoprotein complex that adds telomeric repeat DNA sequences to the chromosome ends, thus maintaining the complex functions of the telomeres. The enzyme activity is upregulated in most malignant tumors. Telomerase activity is usually measured with the telomere repeat amplification protocol (TRAP), a tissue-and cell extract–based assay, first described in 1994.1,2 However, when the TRAP assay has been applied to effusions the results have been contradictory and the utility of telomerase as a marker of malignancy in effusions has been questioned.3–12 Ohyashiki and colleagues13 further developed the TRAP assay to an in situ method, performed on cytospins from fresh cytologic material. When we previously applied TRAP in situ on effusions we achieved high sensitivity figures for malignancy.14,15 A small proportion of mesothelial cell populations were positive, but this reactivity was usually weaker than in malignant cells. Weak or moderate activity was also found in lymphocytes but this presented no problem because the lymphocytes could be identified morphologically.15
In the present study we have compared results from the conventional and the in situ techniques on a series of effusions to elucidate the reasons for the discrepancies of the 2 methods.
MATERIALS AND METHODS
The study comprises 23 effusions from the serous cavities (18 pleural, 5 ascitic fluids). Eighteen of the cases were included in a previous study of TRAP in situ activity in malignant and benign effusions.15 The specimens were obtained and used in agreement with the Swedish legislation regarding biologic material originating from patients and with institutional ethical guidelines. The fluids had been submitted fresh to the Department of Pathology and Cytology, Malmö University Hospital (Malmö, Sweden) from August 2003 through January 2005. The basic materials for the study were the fluids that had been allotted for routine diagnosis to one of us (A.D.). The cytologic diagnoses were based on morphologic characteristics in Giemsa- and hematoxylin-eosin–stained smears, prepared by spinning 20 mL of fluid at 1990g for 10 minutes in a Hettich-Rotica centrifuge (Hettich Zentrifugen, Tuttlingen, Germany), removing the supernatant, and smearing the cell pellet onto slides. All fluids were also immunostained with our standard panel (carcinoembryonic antigen, BerEp4, epithelial membrane antigen, Sialyl-Tn, and cytokeratin 5) for distinguishing between mesothelioma and adenocarcinoma in effusions, based on the results of previous research.16 Analysis of the hyaluronan content was performed with a previously described high-performance liquid chromatography–based method; a hyaluronan value of greater than 75 μg/L was considered positive for mesothelioma, according to previous findings.17 When material had been secured for routine diagnostic purposes, any remaining fluid was used for TRAP in situ and TRAP assay based on cell lysates.
The final diagnoses were based on all available histopathologic, cytopathologic, and clinical information in each case.
TRAP Assay Based on Whole Cell Lysate (TRAP Enzyme-Linked Immunosorbent Assay)
Telomerase activity of whole cell lysate was measured by a Telo TAGGG telomerase PCR ELISA PLUS kit (Roche Diagnostics GmbH, Penzberg, Germany). Cell extract was prepared from frozen cell pellets stored at −80°C by adding 100 μL of a lysis reagent according to the supplier's instruction.
For detection of telomerase activity, first 1 μg and then 0.1 μg of cell lysates were used for each assay, and the subsequent procedure was performed according to the supplier's instruction. For quantification of telomerase activity, we modified the procedure to exclude the presence of polymerase chain reaction (PCR) inhibitors by the adding of phenol/chloroform isoamyl alcohol (CIA) extraction after TS extension. Briefly, 2 μg of cell lysate was incubated with reaction mixture containing telomerase substrate at 25°C for 30 minutes. After primer elongation, template DNA was purified by phenol/CIA extraction, before the performance of PCR. The PCR conditions were 30 cycles at 94°C for 30 seconds, 50°C for 30 seconds, and 72°C for 90 seconds. Hybridization and enzyme-linked immunosorbent assay (ELISA) of PCR products were done according to the supplier's instruction.
The relative telomerase activity (RTA) within different samples was calculated using the following formula:
where Asample = absorbance of sample; Asample,IS = absorbance of internal standard of sample; ATS8 = absorbance of control template; and ATS8,IS = absorbance of internal standard of control template.
TRAP In Situ
The in situ TRAP method has previously been described in detail.13–15
Ten milliliters of the fresh fluid was spun (570g, 5 minutes) in a Hettich-Rotica centrifuge, the pellet was resuspended in hypotonic solution (10mM Tris, 10mM KCl, 1mM MgCl2) and carefully mixed, and hypertonic solution (274mM NaCl, 5.4mM KCl, 16.2mM Na2HPO4, 2.9mM KH2PO4) was added. The cell suspension was cytospun (Cytospin, Thermo Fisher Scientific, Inc, Waltham, Mass) (18g, 3 minutes) onto nonfluorescent slides and dried with cold air. Twenty-five milliliters of a solution containing 20mM Tris-HCl (pH 8.3), 1.5mM MgCl2, 63mM KCl, 0.05% Tween 20, 1mM EGTA (ethylene glycol-bis[β-aminoethyl ether]-N,N,N′,N′-tetraacetic acid), 50mM dNTPs, 1 μg of T4 gene 32 protein, bovine serum albumin (0.1 mg/mL), 2 U of Taq DNA polymerase, and 10 pM fluorescein isothiocyanate–labeled TS forward primer (5′-AAT CCG TCG AGC AGA GTT-3′) were added to a chamber frame (65 μL) that had been placed over each specimen to hold the reaction solutions, and the slides were incubated for 30 minutes at 22°C. After TS extension, 25 μL of the same solution but with 10 pmol of fluorescein isothiocyanate–labeled CX reverse primer (5′-CCC TTA CCC TTA CCC TTA CCC TTA-3′) instead of TS forward primer were added. The reaction chamber was sealed with a cover slip and heated to 90°C for 1.5 minutes to inactivate telomerase. The telomerase reaction product was then amplified using a PTC-100 Programmable Thermal Controller (MJ Research, Inc, Bio-Rad, Calif).
The PCR conditions were 30 cycles at 94°C for 30 seconds, 50°C for 30 seconds, and 72°C for 90 seconds. The chamber frames and top liners were removed completely, and the slides were rinsed in room tempered PBS buffer and then sealed with a cover glass using DAKO Fluorescent mounting medium (DAKO, Glostrup, Denmark).
For each specimen the following control slides were included: 1 slide with all reagents except Taq polymerase, 1 with all reagents except the primers, and 1 with all reagents added but not subjected to PCR. Further, 1 Cytospin slide from each specimen was Giemsa stained for morphologic evaluation.
The TRAP in situ slides were examined with a microscope (Olympus BX51, Tokyo, Japan) allowing alternation between fluorescence (with a B-filter) and phase contrast (U Plane Fluorite ×40 phase contrast objective) to enable the identification also of telomerase-negative cells. Phase contrast microscopy was used to identify the malignant cells found in the routine stained smears. Nuclear fluorescence only or nuclear fluorescence stronger than cytoplasmic fluorescence was considered positive.
The intensity was assigned to 3 groups, weak (1), moderate (2), and strong (3), and an attempt was made to estimate whether all or only part of the cell populations identified showed reactivity. All evaluations were made by authors N.Z. and M.H. and by A.D. independently, to avoid bias, in a blinded fashion without knowledge of the morphologically based diagnoses. Discrepant cases were reevaluated by these 3 authors together, and in all cases consensus was achieved.
Morphologic Evaluation and Estimation of Expected Extract Activity Based on TRAP In Situ Reactivity
Air-dried, Giemsa-stained Cytospin slides were used to evaluate the cell content of the specimens. A semiquantitative evaluation of the amount of mesothelial cells, malignant cells, macrophages, lymphocytes, and granulocytes was made and their proportions assessed. The cell types were ranked according to the proportions in which they occurred in the specimen. If malignant cells were found, their proportion of the whole cell content was roughly estimated in percentage of the total cell content.
Based on the amount of lymphocytes, the proportion of malignant cells, and the intensity of TRAP in situ reactivity, an estimate (low, medium, high) of the probable total telomerase activity in the specimen was made (TRAP estimate). Low activity was expected for cases with few lymphocytes and either less than 20% malignant cells with moderate TRAP in situ reactivity or up to 50% malignant cells with weak reactivity. High activity was expected for cases with more than 85% malignant cells with high TRAP in situ reactivity. For cases with all other combinations of the amount of lymphocytes, proportion of malignant cells, and the intensity of TRAP in situ reactivity, medium activity was expected.
Table 1 shows an overview of the TRAP assay results performed on whole cell lysate and TRAP in situ. In TRAP in situ, if all reactivity intensities are considered positive, the sensitivity for malignancy is 100% and the specificity 71%. If cases with weak TRAP in situ reactivity are also assigned to the negative group, the specificity increases to 100% with remaining 100% sensitivity.
To facilitate the comparison with other extract-based studies attempts to define an optimal cutoff TRAP ELISA value for malignancy were made. When the relative TRAP value 37 was chosen a sensitivity of 100% (14/14) was achieved, however, at the cost of an unacceptably low specificity, 29% (2/7). A cutoff value of 42 (41.3) reduced the sensitivity to 79% (11/14) but gave a specificity of 71% (5/7). These figures are similar to those reported by others (Table 2). Detailed results for all cases are given in Table 3.
TRAP Assay Based on Whole Cell Lysate (TRAP ELISA)
There was considerable overlap between cell lysate– based TRAP assay values in benign (43 ± 9, mean ± SD) and malignant (48 ± 9, mean ± SD) cases.
When telomerase activity first was analyzed using 1 μg of cell lysate, enzyme activity was detected in only 13 of 23 samples. However, no internal standard signal was detected in 18 of 23 samples, indicating the presence of PCR inhibitor(s) in the cell extracts. Thus, the possibility of false-negative results due to PCR inhibition could not be ruled out. For this reason, we analyzed telomerase activity in diluted samples (0.1 μg of cell lysate) giving detectable telomerase activity in all samples, but we still failed to detect the internal standard in 4 of 23 samples (cases 4, 6, 8, and 18; for details see Table 1). When phenol/CIA extraction after TS extension was added to the procedure for the diluted samples, the internal standards were detected in all 23 samples, allowing the quantification of telomerase in all cases.
TRAP In Situ
In 13 of 14 malignant effusions, moderate or strong reactivity was found in the malignant cells (Table 1). One mesothelioma case was negative (case 8), and in this specific case few malignant cells were detected in a heavily blood-contaminated sample. Probably no malignant cells were present on the TRAP in situ slides from this case. In 5 of 7 benign cases there was no reactivity in mesothelial cells (Table 1). In 2 cases (6 and 7) weak reactivity was found in mesothelial cells (in one of the cases reactivity was seen only in one single cell).
Two cases for which no definite diagnosis was established showed moderate and strong reactivity, respectively. In all effusions in which lymphocytes were present, weak or moderate activity was seen in at least part of the lymphocyte population.
Morphologic Evaluation and Correlation Between Conventional TRAP Assay and TRAP In Situ Estimates of Expected Activity
The proportion of different cell types varied considerably in all types of effusions. In the malignant fluids the proportion of tumor cells varied from 5% to 93%. Table 1 shows the cell content and the TRAP estimate for all cases. Figures 1 and 2, showing examples of 2 mesotheliomas, illustrate how the proportion of malignant cells is reflected in the TRAP values. Both mesotheliomas showed strong TRAP in situ reactivity but had high (case 13) and low (case 12) proportions of malignant cells, respectively, and correspondingly high and moderate TRAP estimates.
All but one benign case (case 5) had fairly low relative TRAP values. The exception was a fluid containing a very large amount of neutrophil leukocytes. The patient had pneumonia, the effusion disappeared after treatment, and there has been no relapse. One mesothelioma (case 8) had a rather high TRAP ELISA value in spite of a low TRAP estimate. The fluid was heavily blood stained and the TRAP in situ preparations contained only blood cells. One mesothelioma (case 9) had a moderate TRAP estimate value in spite of a low proportion of malignant cells. The lymphocytes were more numerous than the malignant cells.
Adenocarcinoma effusions with few malignant cells and only moderate fluorescence intensity in TRAP in situ had the lowest values of relative telomerase activity in TRAP ELISA.
Whereas the correlation between the TRAP in situ results and morphologic diagnoses was good there was a considerable overlap between malignant and benign cases for the TRAP ELISA assay performed on whole cell lysates. The TRAP ELISA results are in accordance with those previously reported by others for effusions, as summarized in Table 3.3,4,6–12 There may be several reasons for the poor correlation to malignancy, such as PCR inhibitors, poor sample preservation, extraction problems, hemorrhagic samples, and the cellular composition in the specimen and heterogeneity in telomerase-positive cells. We have discussed these issues in detail in a previous publication.15
To elucidate the contribution of the cellular composition and the intensity of telomerase reactivity to the discrepant results in the present series we tried to estimate the expected TRAP ELISA reactivity in each sample. When this estimate was compared with the TRAP results the discrepancy was reduced. This was true especially for adenocarcinomas, in which low telomerase values were seen in specimens with few malignant cells in relation to the total cell content and moderate telomerase reactivity in the malignant cells.
All but two effusions contained varying amounts of lymphocytes. With the TRAP in situ method we have previously found that at least a fraction of lymphocytes in effusions always show reactivity that may be either weak or moderate. Because upregulation of telomerase activity has been observed in activated lymphocytes, it can be assumed that lymphocytes contribute to the telomerase values obtained with the TRAP ELISA assay, although no exact correlation can be expected.18,19 Several investigators have reported telomerase activity in benign effusion from patients with tuberculosis3,8,9 and it is well known such effusions often contain numerous lymphocytes. Braunschweig and colleagues11 reported telomerase activity in 6 cytologically benign effusions and 2 of these were rich in lymphocytes.
One of the mesothelioma effusions in our series contained very large amounts of lymphocytes but relatively few malignant cells (case 9). The TRAP ELISA value was high, which supports a contribution due to lymphocytes. In the adenocarcinomas, on the other hand, the malignant cells always constituted the largest part of the cell content except in 2 effusions, in which macrophages dominated. In these cases (14, 15) both malignant cells and lymphocytes were few, which was in accordance with the low telomerase value.
Two cases had unexpectedly high telomerase activity levels in conventional TRAP compared with the estimates. Both were TRAP in situ negative. One effusion came from a mesothelioma patient but the Giemsa-stained preparations contained only peripheral blood and no malignant cells (case 8). It cannot be ruled out, however, that the portion taken for TRAP assay performed on whole cell lysate contained malignant cells. The other effusion came from a patient with pneumonia and contained a predominance of granulocytes (case 5). This finding was unexpected, because it is known that granulocytes are telomerase activity negative and also contain RNAses inhibiting telomerase activity.19
The presence of PCR inhibitors is a potential problem with the lysate-based method. In 4 samples internal signals could be achieved only after phenol/CIA extraction. This was a serious drawback for TRAP assay performed on whole cell lysate. For the TRAP in situ assay the cells are immediately fixed onto slides and may be protected from interfering substances released from cells. This may be true especially for effusions, because these often contain many different types of cells. Mesothelial cells play an important role in several biologic processes and are known to produce numerous biologically active substances.20 Thus, it cannot be ruled out that the severe inhibition may at least be due to some extent to the presence of substances produced by mesothelial cells. The immediate fixation of the cells onto slides may also protect the sample from RNA degradation. As TRAP ELISA is an extract-based method, negative results may be due to poor DNA integrity.11
However, in our study, telomerase activity could be demonstrated in all samples and the cell composition and reactivity intensity in TRAP in situ could explain the variation in relative telomerase activity. Thus, poor RNA integrity did not seem to be a significant problem in our study. The extract-based TRAP has the advantage of being easier to perform and is readily quantified. As we have shown, the quantitative outcome will crucially depend on the number of malignant cells present and on the amount of weakly telomerase-expressing nonmalignant cells, which through their number may contribute significantly to the total reactivity. In principle, it would be possible to correct the reading by subtracting a background value based on an independently determined proportion of lymphocytes and mesothelial cells in the sample and their relative telomerase activity and establish diagnostic thresholds with better specificity. However, to our knowledge, the necessary quantitative study of the level and variability of the weak telomerase expression in lymphocytes and mesothelial cells in effusions has not been performed yet.
In the in situ assay, lymphocytes can be identified and their telomerase thus not taken into account. The number of telomerase active cells and their activity can be estimated independently of each other. Even few highly active cells are indicative of malignancy, but weak activity still poses a problem. We are attempting a quantification of in situ activity that would allow one to fix a threshold value with high specificity.
The TRAP in situ results correlated well to final diagnoses, whereas TRAP assay based on whole cell lysate did not differentiate between malignant and benign cases. When an optimal cutoff value for the relative TRAP values was applied, sensitivity and specificity were similar to those reported by others for effusions. The proportion of positive cells and their fluorescence intensity in TRAP in situ explain some of the discrepancies. Lymphocytes may contribute to falsely high values in some cases. If the background activity from mesothelial cells and lymphocytes are considered in the interpretation of TRAP ELISA, a similar comparison can be made between cytology and the TRAP ELISA results and the diagnostic utility of the method can be improved.
This study was supported by grants from The Malm;auo University Cancer Foundation, The Swedish Cancer Society, and LSHC-CT-2004-502943 Mol Cancer Med from the European Union.
The authors have no relevant financial interest in the products or companies described in this article.
Reprints: Nooreldin Zendehrokh, CT, PhD, Lund University, Laboratory Medicine UMAS, Malmö, na 20501 Sweden (Nooreldin.Zendehrokh@med.lu.se)