Improving the cytological diagnosis of high‐grade serous carcinoma in ascites with a panel of complementary biomarkers in cell blocks

Precise cytological diagnosis of pelvic high‐grade serous carcinoma (HGSC) in ascites is important for tumour staging, therapeutic decision‐making and prognostic evaluation. However, it can often be difficult to distinguish metastatic HGSC cells from reactive mesothelial cells based on morphology alone. Immunocytochemical analysis of ascites cell blocks has been used to obtain accurate diagnosis and provide a reliable basis for treatment decisions in the clinic. This study was performed to determine whether a panel of antibodies is necessary to achieve high specificity and sensitivity for the identification of HGSC cells.


| INTRODUCTION
Epithelial ovarian cancer (EOC) is the fifth-most common cause of cancer death in women and a common cause of death due to gynaecological tumours. 1 High-grade serous carcinoma (HGSC) is the most common subtype encountered in patients with EOC. 2 Being largely asymptomatic, over 70% of patients are diagnosed at an advanced stage of disease (stage III/IV) with metastasis throughout the peritoneal cavity and a substantial amount of ascites. 3 The primary cytology of ascites is an important for diagnosis, therapeutic decision and prognostic prediction. The results of secondary cytology after treatment are also important independent prognostic indicators that are highly correlated with optimal surgery outcomes, recurrence and overall survival rate. 4 However, an increasing number of studies have shown that the morphological examination of cytological samples is not a highly sensitive diagnostic tool to distinguish metastatic adenocarcinoma from reactive mesothelial cells in ascites. Immunocytochemistry is commonly used in cytopathological diagnosis. This study was performed to establish whether a panel of antibodies, including two markers for HGSC and two markers for mesothelial cells, can increase the rate of detection of metastatic tumour cells.

| Case selection
All the samples were selected from the Department of Pathology at

| Immunohistochemical staining
Cell blocks were fixed in 10% neutral-buffered formalin and processed in a standard tissue processor. Immunocytochemistry was performed on 4-lm paraffin sections from cell blocks corresponding to each case. The sections were deparaffinized in xylene and rehydrated through a graded alcohol series. The sections were treated with 3% H 2 O 2 for 10 min to inactivate endogenous peroxidase.
T A B L E 1 Summary of sensitivity and specificity of epithelial markers in high-grade serous carcinoma ascites cell blocks

| Data analysis
We used the results from haematoxylin and eosin (HE) staining as gold standard and those of immunostained markers as new findings to construct a 4 9 4 table to calculate both sensitivity and specificity of each marker as shown in Tables 1 and 2

| Ber-EP4
In total, 70 cases of ascites cell blocks were stained for Ber-EP4.

| E-cadherin
In total, 29 cases of ascites cell blocks were stained for E-cadherin.
Among 20 ascites-positive cases based on HE staining, 19 were positive for E-cadherin. Therefore, the sensitivity of E-cadherin was 95% (19/20). Nevertheless, six ascites-negative cases based on HE staining also showed E-cadherin expression in mesothelial cells.

| Calretinin
Twenty-nine available cases of ascites cell blocks were stained for calretinin. Mesothelial cells in eight ascites-negative cases based on | 249 HE staining were positive for calretinin. The sensitivity of calretinin was thus 88.9% (8/9). Two ascites-positive cases also exhibited calretinin-positivity in tumour cells; the specificity of calretinin was therefore 90% (18/20; Table 2).  (Table 1). In addition, because of the reduction in the false-positive rate of HBME-1 and false-negative rate of calretinin, the sensitivity and specificity of HBME-1 and calretinin also increased to 100% and 90%, respectively ( Table 2).

| Application of the panel of complementary biomarkers in ascites cytology
These four biomarkers were used to construct a panel of complementary markers for the differential diagnosis of HGSC based on ascites cytology. As shown in Figure 1, However, because of its high specificity but relatively poor sensitivity, Ber-EP4 staining appeared falsely negative in some cases (Figure 2). Nevertheless, these Ber-EP4-negative cells were positive Ecadherin, negative for calretinin, and weakly positive for HBME-1.
By comprehensive analyses, these cells were confirmed as metastatic epithelial tumour cells. By contrast, because of its high sensitivity and poor specificity, HBME-1 gave false-positive results in some cases ( Figure 3). In one preoperative case, Ber-Ep4 and E-cadherin were positive in the membrane of some cells. However, some parts of the membrane in these cells were also positive for HBME-1.
Because these cells were negative for calretinin expression, HBME-1 probably cross-reacted with tumour cells and thus gave rise to falsepositive results in this case.
Among 16 patients with preoperative ascites, three simultaneously had pleural effusion. Among these cases, 15 (93.7%, 15/16) were confirmed to be positive for tumour cells in ascites. The immunocytochemical staining results are summarized in Table 3.

| DISCUSSION
Precise cytological diagnosis of pelvic HGSC from ascites is important for tumour staging and therapeutic decision-making, especially for patients who need accurate tumour typing by preoperative and the specificity of Ber-EP4 in adenocarcinomas was as high as 83%-100%. 8,9 The antibody against the mesothelial cell marker HBME-1, which has been isolated from a human malignant epithelial mesothelioma cell suspension, can bind to an antigen on the surface of mesothelial cells. 7 HBME-1 has been reported to be expressed in 76%-100% of mesothelioma cells, 77%-100% of reactive mesothelial cells, and 15%-100% of adenocarcinoma cells. [7][8][9][10] In this study, the sensitivity and specificity was 85.7% and 82.1% for Ber-EP4 and 100% and 68.3% for HBME-1, respectively. Thus, our results are consistent with other reports.
In view of the poor sensitivity and a relatively low specificity of Ber-EP4 and its cross-reaction with HBME-1, a panel of antibodies is needed for more precise diagnosis. Therefore, we examined another marker of epithelial cells, E-cadherin, because this marker was expressed in 72%-100% metastatic adenocarcinomas in the serous effusions and because its specificity ranged from 54% to 100%. 11 In addition, the expression of calretinin, an intercellular calcium-binding protein belonging to the troponin C superfamily, was also examined in our study. The cytological examination of serous effusions showed that the rate of positivity for calretinin in mesothelial cells ranged from 58% to 100%, with a specificity of approximately 91%-100%. 12 Based on our study, the sensitivity and specificity were 95% and 33.3% and 88.9% for E-cadherin and 90% for calretinin, respectively. Other than the four major markers used in our study, many immunostaining markers such as MOC-31, EMA, WT-1, D2-40, CK5/6, CA199, CEA, mesothelin and thrombomodulin have been used for the differential diagnosis of adenocarcinoma cells and mesothelial cells. 7,9,10 However, the sensitivity and specificity of these markers varies from 21% to 100% when used alone. 7