Department of Pathology, Fukuoka University Hospital and School of Medicine, Fukuoka, Japan
Corresponding author: Kazuki Nabeshima, MD, PhD, Department of Pathology, Fukuoka University Hospital and School of Medicine, Fukuoka 814-0180, Japan; Fax: (011) 81-92-863-8383; firstname.lastname@example.org
We thank Ms. A. Oogami and Ms. K. Yano for their skillful help in p16 FISH and Dr. Miyake, Department of Preventive Medicine and Public Health, Fukuoka University School of Medicine, for his kind suggestions in statistical analysis.
In malignant pleural mesothelioma (MPM), most patients first present with pleural effusion; thus, cytologic analysis is the primary diagnostic approach. However, the cytologic distinction between MPM and reactive mesothelial cells (RMCs) in effusions can be extremely difficult due to the lack of both well-established immunocytochemical markers and definite cytological criteria for MPM. Moreover, the existence of both MPM cells and RMCs in effusions from the same patient makes the differentiation even more challenging. Homozygous deletion of the 9p21 locus, the site of the cyclin-dependent kinase inhibitor 2A/p16 (CDKN2A/p16) gene, frequently occurs in MPM but has never been reported in RMCs. The aim of this study was to define the cytomorphological characteristics of MPM cells, identified by the presence of 9p21 homozygous deletion by fluorescence in situ hybridization (FISH).
For this purpose, cells on smear preparations were recorded using a virtual microscope system and were subjected to FISH analysis. Thereafter, 9p21 homozygous deletion-positive cells were identified in the recorded virtual slides, followed by analysis of their morphological characteristics.
Mesothelioma cells positive for the 9p21 homozygous deletion exhibited significantly more frequent cell-in-cell engulfment, multinucleation (more than 2 nuclei), and larger multicellular clusters composed of more than 10 cells than did 9p21 deletion-negative RMCs. Possible cutoff values are also proposed for these morphological markers to differentiate MPM cells from RMCs.
The incidence of malignant pleural mesothelioma (MPM) is increasing worldwide and is expected to reach a peak in approximately 2020. MPM is an aggressive tumor associated with exposure to asbestos, with a median survival time after diagnosis of 9 to 17 months. The poor prognosis is related, at least in part, to difficulties in diagnosis in the early stage of the disease. Imaging techniques have not proven helpful in the diagnosis of MPM. Because 54% to 89% of patients first present with pleural effusion, cytologic analysis is the primary diagnostic approach for most patients.[3-5] A definitive discrimination of MPM from pleural metastases derived from other primary carcinomas can be made in most cases by using an appropriate panel of immunocytochemical markers. However, cytologic distinction between MPM and reactive mesothelial cells (RMCs) in effusions can be impossible sometimes due to the lack of both well-established immunocytochemical markers and definitive cytological criteria for MPM. Moreover, the copresence of mesothelioma cells and RMCs within effusions in the same patient makes the differentiation difficult.
Deletion involving the 9p21 locus, the site of the cyclin-dependent kinase inhibitor 2A/p16 gene (CDKN2A/p16), is common in mesothelioma.[5-8] A homozygous 9p21 deletion has been reported in 22% to 74% of mesotheliomas. The 9p21 deletion, detected by fluorescence in situ hybridization (FISH) in cytological preparations and tissue sections, has been suggested as a potentially useful clinical assay for the diagnosis of MPM.[5-11] These same studies have also confirmed the specificity of 9p21 deletion in MPM by noting its absence in cytologically and histologically negative specimens (ie, in RMCs).[5, 7, 8, 10-13]
The purpose of this study was to characterize the morphological features of definite mesothelioma cells, classified as exhibiting a FISH-defined homozygous deletion of the 9p21 locus. For this purpose, cells on all smear preparations were first recorded using a virtual microscope system, and then the smears were subjected to FISH analysis. Next, 9p21 homozygous deletion-positive cells were identified in the recorded virtual slides, and their morphological features were characterized. The results showed that cell-in-cell engulfment (including hump formation), multinucleation, and multicellular clusters are specific characteristic features of mesothelioma cells.
MATERIALS AND METHODS
This study included cytological preparations obtained from pleural effusions in 15 patients with MPM (14 males and 1 female; mean age, 68.5 years; range, 49-85 years) and in 20 patients with RMCs associated with tuberculosis, pneumonia, cardiovascular disease, lung cancer, and bullae (12 males and 8 females; mean age, 70.6 years; range, 36-95 years). All 35 cases were derived from the pleural lesion files of the Department of Pathology, Fukuoka University Hospital, Fukuoka, Japan, including consultation cases. Anonymous use of redundant tissues and cells is part of the standard treatment agreement with patients in our hospital when no objection is expressed. Furthermore, the study protocol was approved by the Ethics Committee of Fukuoka University. The diagnosis of all 15 MPM effusion cases had been established histologically and immunohistochemically. All MPM subtypes were epithelioid according to the 2003 classification by the World Health Organization.
Pretreatment of Specimens and FISH Assay
All smear preparations, ordinarily fixed in 95% ethanol and stained according to Papanicolaou, were morphologically analyzed using a virtual microscope system (NanoZoomer 2.0-HT; Hamamatsu Photonics, Hamamatsu, Japan) and then subjected to FISH analysis using dual-color probes; a Spectrum Green-labeled chromosome 9 centromeric probe and a Spectrum-Orange labeled, locus-specific CDKN2A (p16) probe (Vysis LSI p16/CEP 9 probe; Abbott Japan, Tokyo, Japan). Briefly, smear preparations were treated in xylene overnight to remove the mounting medium, followed by rehydration in descending alcohol dilutions and fixation in Carnoy's solution at room temperature (RT) for 15 minutes. This was followed by treatment with 2× saline-sodium citrate (SSC) containing 0.3% Tween 20 (Sigma, St Louis, Mo) at 37 °C for 1 to 12 hours, and then with pretreatment solution (Histology FISH Accessory Kit; Dako, Carpinteria, Calif; 20× dilution) at 85 °C for 15 minutes, and digestion with pepsin solution (DAKO, 5× dilution) at 37 °C for 1 to 2 minutes. After refixation in 10% buffered formalin at RT for 5 minutes, the smears were treated in 2× SSC containing 0.3% Tween 20 at 42 °C for 30 minutes, dehydrated in ethanol, and dried, followed by addition of the 2 probes. Both the probes and smears were denatured at 85 °C for 5 minutes in the probe solution included with the Kit (Abbott Japan) followed by hybridization at 37 °C for 48 hours in ThermoBrite (Abbott Japan).
The smears were washed in 2× SSC containing 0.3% Tween 20 at 72 °C for 3 to 5 minutes and in 2× SSC containing 0.1% Tween 20 at RT for 3 to 5 minutes. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI)/antifade (Vector Laboratories, Burlingame, Calif). Analyses were performed using a fluorescence microscope (Axio Imager Z1; Carl Zeiss Microimaging, Jena, Germany) and Isis analysis system (Metasystems, Altlussheim, Germany) equipped with filter sets with single and dual band excitors for Spectrum Green, Spectrum Orange, and DAPI (ultraviolet wavelength = 360 nm). Lymphocytes in each smear served as internal controls and showed 2 signals per FISH probe. Smear preparations with cultured mesothelioma cells (MESO-4) previously identified as carrying a homozygous p16 deletion served as a positive control. Homozygous deletion was defined as lack of both 9p21 signals. Heterozygous deletion was assumed when only one 9p21 signal was present, or when the total number of 9p21 signals did not exceed half the total number of the centromeric signals. At least 100 cells were scored in each case.
According to the results of FISH analysis using RMC smears, a FISH result was considered positive if at least one of the following criteria was met: 1) homozygous deletion of 9p21 was identified in > 10% of mesothelial cells, and/or 2) heterozygous deletion of 9p21was identified in > 30% of mesothelial cells. The areas with homozygous deletion-positive cells selected on a FISH-treated smear were identified on the Papanicolaou-stained smear of MPM recorded by the virtual microscope system, and positive cells were marked in the system. Similarly, RMCs with a FISH-confirmed 9p21 normal pattern were marked on the Papanicolaou-stained smear of nonmesothelioma cases. These enabled us to analyze the morphological characteristics of homozygous deletion-positive mesothelioma cells and deletion-negative RMCs. The morphometric measurements, including nucleolar surface area expressed in square micrometers (μm2) and nucleolar shape parameter assessing the degree of nucleolar roundness, were performed using analytical software equipped with NanoZoomer 2.0-HT (Hamamatsu Photonics).
Statistical comparison of FISH data and cytological features between MPM and RMCs was performed with the Mann-Whitney U test. A P value < .05 was considered statistically significant. All statistical evaluations were performed with the StatView statistical software for Windows, version 5.0.1 (SAS Institute, Cary, NC).
9p21 FISH in Effusion Smears and Cutoff Values
First, the cutoff values were calculated as the mean percentage ± 3 standard deviations (SDs) of nuclei showing one 9p21 signal for heterozygous deletion, and that of nuclei showing loss of both 9p21 signals for homozygous deletion as described. Among cases with RMCs, a homozygous deletion pattern was observed in only 0% to 3.6% (mean, 1.3%) and a heterozygous deletion pattern in 0.9% to 33% (mean, 8.4%). A normal 9p21 signal pattern was most frequently noted in 65.1% to 98.6% (mean, 90.3%) of cases (Fig. 1). Based on these results, a cutoff value of >10% for nuclei with a homozygous deletion pattern was set for homozygous deletion (although the actual value was 4.3%, >10% was used to exclude the possibility of pseudopositives). A cutoff value of > 30% was set for a heterozygous deletion. Cases with either positive homozygous or heterozygous deletion were defined as 9p21 deletion-positive. Based on these criteria, all 15 MPM cases were 9p21 deletion-positive, whereas none of the RMC cases were positive (sensitivity of 100%, specificity of 100%). Of the 15 MPM cases, 12 were positive for a homozygous deletion alone and 3 were positive for both homozygous and heterozygous deletions with a predominantly heterozygous pattern (homozygous deletion = 10.6% plus heterozygous deletion = 62.9%; 25% plus 62.0%, respectively; and 28.6% plus 45.0%, respectively, for the 3 cases). The mean percentage of nuclei with a homozygous deletion was significantly higher in MPM cases (mean, 77.8%; range, 10.6%-98.4%) than in RMC cases. The percentage of nuclei with a heterozygous deletion was 16.2% (range, 0.8%-62.9%). A normal pattern was seen in only 0% to 26.5% (mean, 6.0%).
Morphology of 9p21 Homozygous Deletion-Positive Mesothelioma Cells
In MPM smears, which possibly included both mesothelioma cells and RMCs, mesothelial cells harboring a 9p21 homozygous deletion were interpreted as mesothelioma cells given that genomic alterations are a hallmark of malignant cells. Homozygous deletion-positive mesothelioma cells were marked and selected on the images recorded using the virtual microscope system to define the morphological characteristics. Similarly, 9p21 deletion-negative RMCs were selected from the recorded images of non-neoplastic cases with RMCs. Mesothelial cells harboring a 9p21 homozygous deletion exhibited significantly more frequent cell-in-cell engulfment (with or without hump-like appearance) than RMCs (P < .05, Mann-Whitney U test) (Fig. 2A,B; Fig. 3). Moreover, 11 of 15 cases (73%) of MPM showed cell-in-cell engulfment at a higher rate than the cutoff value (14.3, calculated by the mean ± 3 SDs in RMCs).
Multinucleation with 2 or more nuclei was also significantly more frequent in homozygous deletion-positive mesothelioma cells (P < .05, Mann-Whitney U test) (Fig. 2C,D; Fig. 3). However, binucleation was less discriminating between reactive mesothelial and MPM cases: even among mesothelioma cases, the binucleation rates were below the cutoff value (49.1%). In contrast, multinucleation with more than 2 nuclei was more valuable in the discrimination: 14 of 15 MPM cases (93%) showed a multinucleation rate over the cutoff value of 9.3%, whereas all RMC cases remained below the cutoff value. Three or more nucleoli were seen significantly more often in RMCs than in mesothelioma cells (Fig. 4A). Mesothelioma cells often had one larger, rounder, more expanded nucleolus compared with that of RMCs (P < .05, Mann-Whitney U test) (Fig. 4B,C). However, the differences were not large enough to set useful cutoff values for differentiation.
Multicellular ball or berry-like clusters of mesothelial cells were also more often found in deletion-positive mesothelioma cells. However, these multicellular clusters were 3-dimensional with many piled-up cells, especially in the middle, and thus determination of 9p21 homozygous deletion by FISH could be achieved only at the periphery of the clusters (Fig. 2E,F). Based on this limitation, statistical analysis of the multicellular clusters was performed only in 9p21 homozygous deletion-predominant MPM cases, in which more than 90% of cells were homozygous deletion-positive (n = 10). The multicellular 3-dimensional clusters composed of more than 10 or 100 cells were significantly characteristic for deletion-positive mesothelioma cells but not RMCs (P < .05, Mann-Whitney U test) (Fig. 5). Eight of 10 MPM cases (80%) with multicellular clusters composed of 10 to 99 cells showed a cluster rate over the cutoff value of 5.5%, whereas the cluster rates of all reactive cases were below the cutoff value. RMCs more often formed smaller monolayer cell clusters.
In this study, cytologic preparations from pleural effusions were subjected to virtual microscope analysis and 9p21 FISH to characterize the morphological features of mesothelioma cells identified by the presence of 9p21 homozygous deletion. Compared with 9p21 deletion-negative RMCs, 9p21 homozygous deletion-positive mesothelioma cells exhibited significantly more frequent cell-in-cell engulfment, more frequent multinucleation (with more than 2 nuclei), and larger multicellular berry-like clusters composed of more than 10 cells. In addition, mesothelioma cells tended to have one large, round nucleolus. These features were useful for cytological differentiation of mesothelioma cells from RMCs. We also introduced a novel methodology to analyze tumor morphology using a combination of the virtual microscope system and FISH.
Although usual cytologic criteria for malignancy such as nuclear/nucleolar enlargement, irregularity, and hyperchromasia still apply to malignant mesothelioma, these features are often not present, and other cytologic criteria have been reported to be useful. The predictive value of each of 24 cytologic features, previously reported to be of use in the distinction of malignant mesothelioma, adenocarcinoma, and benign mesothelial proliferation in serous effusions, was examined by an Australian group. Four cytologic features were found to differentiate malignant mesothelioma from benign mesothelial proliferation, including nuclear pleomorphism, macronucleoli, and cell-in-cell engulfment, or “clasping.”
Another study found that cell ball formation, cell-in-cell engulfment, and monolayer cell groups, a feature of benign mesothelial proliferation, all aid in differentiation of malignant mesothelioma from RMCs. Although binucleate forms of benign mesothelial cells may be present, the presence of many large multinucleated cells with atypical nuclei is considered a hallmark of the epithelial form of mesothelioma.[15, 18] Moreover, in mesothelioma cells, formation of a hump-like cytoplasmic process is associated with cell-in-cell engulfment. The cytoplasm of the engulfed cells protrudes outside the engulfing cells, exhibiting a hump-like cytoplasmic structure.
Our study showed that berry-like multicellular clusters (predominantly those composed of more than 10 cells), multinucleation with more than 2 nuclei, and cell-in-cell engulfment with or without formation of a hump-like cytoplasmic process were significantly more frequently observed in 9p21 homozygous deletion-positive mesothelioma cells than in RMCs harboring a normal 9p21 locus. Moreover, we proposed tentative cutoff values to favor diagnosis of mesothelioma; 14.3%, 9.3%, and 5.5% for cell-cell engulfment, multinucleation with more than 2 nuclei, and multicellular clusters composed of 10 to 99 cells, respectively. Observed frequencies of the described morphological characteristics above the calculated cutoff values help in distinguishing MPM from RMCs. To make the cutoff values more reliable, however, additional studies with MPM and control cohorts are needed. In addition, analysis on the morphology of heterozygous deletion-positive mesothelioma cells is now under investigation in our laboratory. It is preliminary, but the morphological features of heterozygous deletion-positive tumor cells seem to be similar to those of homozygous deletion-positive MPM cells: for example, cell-in-cell engulfment with or without formation of a cytoplasmic hump was significantly more frequent in both homozygous and heterozygous deletion-positive MPM cells than in RMCs harboring a normal 9p21 locus, and the difference between the homo- and heterozygous deletion-positive MPM cells was insignificant (data not shown).
CDKN2A is present in the 9p21 locus and encodes 2 important cell cycle regulatory proteins: the p16 protein and, in an alternative reading frame, the p14ARF protein. The cyclin-dependent kinase inhibitor p16 blocks the phosphorylation of the RB protein and p14ARF blocks MDM2, resulting in positive regulation of p53. CDKN2A is considered a helpful marker for differentiating benign reactive mesothelium from cells of malignant mesothelioma, because CDKN2A, although essential for normal cell cycle control, is lost in the majority of malignant mesotheliomas. In addition to this study, other recent reports on CDKN2A/p16 FISH in MPM versus reactive mesothelial tissue or cells indicate that approximately 69% (217 of 314 cases; range, 43%-100%) of mesothelioma cases are homozygous deletion-positive, whereas no reactive mesothelial cases (0 of 177 cases) are deletion-positive.[5, 7, 8, 10-13] The specificity of p16 FISH in discriminating MPM from reactive mesothelium was 100% in all studies. In the current study, we analyzed the morphology only in homozygous deletion-positive mesothelial cells present in pleural effusion smears obtained from patients with mesothelioma, because the positive cells could be considered definite mesothelioma cells with high accuracy. Although smears can potentially contain both mesothelioma cells and RMCs, in our protocol, the potential of analysis of RMCs instead of mesothelioma cells was reduced to a minimum. To our knowledge, this study is the first to use the p16 FISH pattern to select mesothelioma and RMCs for morphological differentiation. p16 FISH is a useful ancillary test in the cytologic or histopathological diagnosis of MPM, although it cannot be performed in all hospital laboratories. In this regard, it is essential to analyze the morphological features of definite mesothelioma cells in Papanicolaou preparations.
This work was supported in part by a grant from the Research Center for Advanced Molecular Medicine, Fukuoka University.