Role of Maspin, CK17 and Ki-67 Immunophenotyping in Diagnosing of Pancreatic Ductal Adenocarcinoma in Endoscopic Ultrasound-Guided Fine Needle Aspiration Cytology

Background and study aim: One of the problems in diagnosing pancreatic ductal adenocarcinoma (PDAC) is differentiation between PDAC cells and benign pancreatic tissue cells in cytologic samples. This study aimed to evaluate the usefulness of Maspin, CK17 and Ki-67 immunocytochemistry (ICC) in differentiation between these two groups of cells. Materials and methods: This retrospective study was carried on 80 cases of PDAC and 25 cell blocks of benign pancreatic tissue cells as a control group for evaluation of Maspin, CK17 and Ki-67 ICC. PDAC cases were sampled by endoscopic ultrasound guided fine needle aspiration cytology (EUS-FNAC), while cell blocks of control group were aspirated from benign pancreatic tissues that were obtained from the pancreatic surgically resected specimens. Immunostaining patterns, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy of each antibody as well as possible antibody combined panels of these markers in differentiation between the two groups were evaluated. Results: Positive immunoreactivity for Maspin, CK17 and Ki-67 were 92.5%, 80% and 72.5% in PDAC cases, respectively. In contrast to PDAC cases, all the cell blocks of benign pancreatic tissue cells were negative for these markers. Regarding different panels, combined use of Maspin, CK17 and Ki-67 together as a triple test (at least one of them is positive) achieved the highest sensitivity of 98.8%, specificity of 100%, PPV of 100%, NPV of 96.2% and accuracy of 99% in the differentiation between PDAC and benign pancreatic tissue. Conclusion: Employing this short panel [Maspin, CK17 and Ki-67] is helpful for better differentiation between PDAC and benign pancreatic tissue.


Role of Maspin, CK17 and Ki-67 Immunophenotyping in Diagnosing of Pancreatic Ductal Adenocarcinoma in Endoscopic Ultrasound-Guided Fine Needle Aspiration Cytology
lesions is a safe and efficient procedure and has become a popular approach for obtaining diagnostic pancreatic samples (Kudo et al., 2014).
CK17 is a low-molecular-weight keratin that is normally expressed in myoepithelial, basal cells and subsets of hair shaft epithelia (Chu et al., 2005;Lok et al., 2014). It is also expressed mainly in squamous, basal and transitional cell carcinomas as well as in adenocarcinomas with squamous differentiation (Goldstein and Bassi, 2001;Yang et al., 2012). Few studies had shown that CK17 also might be a useful marker for the diagnosis of pancreaticobiliary adenocarcinomas (Goldstein and Bassi, 2001;Berardi et al., 2013) and separating them from extra-pancreaticobiliary non-mucinous adenocarcinomas (Chu et al., 2005;Sarbia et al., 2007;Yang et al., 2012).
Ki-67 is a nuclear protein that is increased in proliferating cells and present in all phases of the cell cycle, except the resting phase (Jahng et al., 2010). Overexpression is frequently seen in a variety of malignant tumors and is associated with worse prognosis (Stuart-Harris et al., 2008;Viale et al., 2008). A stepwise progression from normal pancreatic epithelium to pancreatic intraepithelial neoplasia (PanIN) and then to frank adenocarcinoma has been correlated with increasing Ki-67 labeling index expression (Klein et al., 2002;Jahng et al., 2010;Karamitopoulou et al., 2010;Kim et al., 2005;Furuhata et al., 2017).
Therefore, this study aimed to evaluate the usefulness of Maspin, CK17 and Ki-67 ICC as separate markers and as different combined panels in differentiation between PDAC cells and benign pancreatic tissue cells.

Materials and Methods
This retrospective study was conducted on 80 cases of PDAC and 25 cell blocks of benign pancreatic tissue cells as a control group in National Cancer Institute (NCI), Cairo University.

EUS-FNAC procedure and preparation of the cases
The aforementioned 80 cases were aspirated by EUS-FNAC that was performed using 22-gauge needle with rapid-on-site evaluation (ROSE) in the Endoscopic Unit. Smears were prepared from the aspirate of the needle pass: at least, one was air-dried and Diff-Quik-stained (used for ROSE) and the other was fixed in 95% Ethyl alcohol for minutes at room temperature. If the obtained sample was non-diagnostic, additional passes were attempted until diagnostic material is obtained. Additional passes or needle rinse material was collected in solution [10% neutral-buffered formalin: alcohol, 1:9] for subsequent preparation of cell block. In addition, some material was collected in SurePath liquid based cytology (LBC) preservative.
All the aspirated material was sent to the Cytology Unit, for preparation, immunostaining and cytological diagnosis. The Ethyl alcohol-fixed slides were stained using Papanicolaou stain, while cell block material was prepared and stained with HandE stain. Whereas Surepath LBC material was prepared and stained using SurePath LBC apparatus.
Regarding the 25 cell blocks of control group, they were prepared from aspirated benign pancreatic tissues that were obtained from the pancreatic surgically resected specimens that were sent to our Pathology Unit (25 Whipple operations for PDAC) and then stained using HandE stain.

Selection of cases
These PDAC cases were retrieved from the database of Cytology Unit, during the period from January 2014 to December 2019. The selected criteria of the cases included unequivocal cytologic reports of PDAC, availability of all studied correlated clinical and radiologic/endoscopic data, availability of formalin-fixed paraffin-embedded cell blocks with adequate tumor cells to perform ICC and no administered chemo-or radiotherapy before FNAC.

Cytologic examination and interpretation
The review of diagnosis and grading of our cases was done according to "The Papanicolaou Society of Cytopathology System for Reporting Pancreaticobiliary Cytology" (Pitman et al., 2014). The diagnosis of the cases was correlated with their final histopathological diagnoses (if present) and radiologic/endoscopic data that included malignant criteria of the pancreatic mass, pathological lymph nodes and/or presence of distant metastases. Also clinical course was considered.

Immunocytochemical (ICC) staining
For immunostaining, Paraffin embedded sections of cell blocks of PDAC cases and the control group were made at 4 microns thickness and mounted on positive charged slides. Immunostaining was done by BenchMark IHC/ISH staining module (Ventana immunostaining system), the following steps occurred automatically: 1) Deparaffinization, 2) Cell conditioning (Standard CC1 application) for 80 minutes, 3) Application of one drop (100µ) of the antibody, 4) Application of cover slip and incubation for 32 minutes, 5) Application of one drop of DAB (counterstain) with Haematoxylin II and incubation for 8 minutes, 6) Application of Bluing Reagent for 4 minutes, 7) Slides were extracted and arranged in racks, 8) Slides were washed in tap water and soap for 5 minutes and then dehydrated in the ascending grades of alcohol for 5 minutes in each container, 9) Slides were cleared in xyline, and then cover slips were applied. Value of Maspin, CK17 and Ki-67 in diagnosing Pancreatic Ductal Adenocarcinoma and standard deviations or medians and/or ranges, as appropriate. Categorical data were summarized as numbers and percentages. Numerical data were explored for normality using Kolmogrov-Smirnov test and Shapiro-Wilk test. Chi square or Fisher's tests was used to compare between independent groups with respect to categorical data. Comparisons between two groups for normally distributed numeric variables were done using the Student's t-test while for non-normally distributed numeric variables, comparisons were done by Mann-Whitney U test. To measure the strength of association between the non-normally distributed measurements, Spearman's correlation coefficients was calculated (r is the correlation coefficient, it ranges from -1 to +1), (+1 indicates positive correlation, -1 indicates negative correlation, 0 indicates no correlation). Receiver operating characteristic curve (ROC curve) was done to determine the best sensitivity, specificity and area under the curve. The accuracy of the test depends on how well the test separates the group being tested into those with and without the disease in question. All tests were two tailed and Probability (p-value) ≤ 0.05 is considered significant.

Results
Our PDAC cases included 51 male and 29 female with male: female ratio of 1.8:1 and a mean age of 60 years. Among the studied cases, recorded reports of CA19-9 serum level were only available for 63 cases and those of CEA serum level were available for 58 cases. High CA19-9 and CEA serum levels were documented in 74.6% and 37.9% of available reports, respectively. The majority of our cases were located at the head of the pancreas, represented 71.2%. Regarding TNM staging groups, the majority of our cases were of high stage; stage III and Appropriate positive and negative controls were included in each run. The selected markers were Maspin (clone BSB-92, Ready to use, Gene Tex), CK17 (clone E3, Ready to use, Dako) and Ki-67 (clone MIB-1, Ready to use, Dako).
We correlated immunostaining results of PDAC cases with some clinico-pathological data.

Statistical analysis
Data management and analysis were performed using Statistical Package for Social Sciences (SPSS) vs. 25. Numerical data were summarized using means IV represented 15% and 43.8%, respectively. Most of cases were moderately differentiated tumors, represented 63.7% (Table 1).

Immunocytochemical findings
The immunostaining results for PDAC cases are summarized in (    (Figure 2 and 3). In contrast to PDAC cases, all the cell blocks of benign pancreatic tissue cells were negative for Maspin, CK17 and Ki-67 (Figure 4), with statistical significant difference in differentiation between the two groups (P < 0.01).
In differentiation of PDAC from benign pancreatic tissue, Maspin showed sensitivity of 92.5%, NPV of 80.6% and accuracy of 94.2%. CK17 sensitivity, NPV and accuracy were of 80%, 61% and 84.8%, respectively. Finally, Ki-67 had sensitivity of 72.5%, NPV of 53.2% and accuracy of 79%. Specificity and PPV were 100% for each marker. Regarding different panels, combined use of the three markers together as a triple test (at least one of them is positive) achieved the highest sensitivity of 98.8 %, NPV of 96.2 % and accuracy of 99 % in the    differentiation between PDAC and benign pancreatic tissue. Whereas, specificity and PPV were 100% for each of the studied panels (Table 3). Among all studied clinico-pathological data, Ki-67-positive PDAC cases only showed statistical significant relation with tumor grade, compared with negative ones. Ki-67 positivity was significantly higher in poorly differentiated PDAC, represented 22 out of 24 cases (91.7%), compared to well and moderately differentiated tumors, represented 3 out of 5 cases (60%) and 33 out of 51 cases (64.7%), respectively. In contrast, both Maspin and CK17 immunostaining were not significantly correlated with all studied clinico-pathological data (Table 4).

Discussion
Differentiation of PDAC, especially low grades, from benign pancreatic tissues is one of the obstacles that facing cytopathologists in the diagnosis of pancreatic lesions in EUS-FNAC (Furhata et al., 2017). In this study we aimed to providing information about a new combination of ICC antibodies (Maspin, CK17 and Ki-67) as an ancillary study Value of Maspin, in diagnosis of EUS-FNAC of PDAC. Although many studies evaluated one or two of these markers in diagnosis of PDAC (Yang et al., 2012;Furuhata et al., 2017;Aksoy-Altinboga et al., 2018), to our best knowledge, we are the first in the literature to test the combination of these three markers together and we are also the first to study the usefulness of CK17 in diagnosing PDAC on cytologic material.
In our study, we found that Maspin, CK17 and Ki-67 were very useful markers in differentiating PDAC from benign pancreatic tissue with a sensitivity of 92.5%, 80% and 72.5%, respectively and the specificity was of 100% for each marker. Among our control group, no one showed positive staining for any of the studied markers.
In the existing work, the expression of Maspin in PDAC cases was almost similar to the majority of studies in literature (>/= 90% of cases) and on the same line, all the benign pancreatic tissues were negative for it (Cao et al., 2007;Berardi et al., 2013;Aksoy-Altinboga et al., 2018). Aksoy-Altinboga et al., (2018) reported that using Maspin in the differential diagnosis of PDAC from benign/ reactive pancreatic tissue achieved diagnostic sensitivity and accuracy of 87.5% and 91.2% respectively, which were slightly lower than what we recorded, while they recorded higher NPV of 89%, whereas specificity and PPV were similar to our findings (100%, each).
Regarding the subcellular localization of Maspin immunostaining positivity, there is wide variation in the literature with lack of standardization. For example, Cao et al., (2007) documented that 87% of cases had only cytoplasmic expression, while the majority of our Maspin positive PDAC cases (77%) had nuclear and cytoplasmic immunostaining. Berardi et al., (2013) and Banias et al., (2019)   expression and its subcellular localization influence both tumor behavior and response to chemotherapy. However few studies in literature investigated the significance of subcellular localization of Maspin in PDAC. Cao et al., (2007) proved that nuclear labeling of Maspin in PDAC is associated with better tumor differentiation although it is not associated with a better prognosis. Berardi et al., (2013) showed that nuclear expression of Maspin in PDAC predicts a poor prognosis.
All studies of CK17 in PDAC come in literature were done on surgically resected specimens, with wide variation in positivity rate. Higher figures were noted by Chu et al., (2005Chu et al., ( ), sarbia et al., (2007 and Yang et al., (2012), represented 83%, 88 % and 92% of PDAC cases, respectively. In contrast, CK17 was only found in 60 % of pancreaticobiliary adenocarcinomas (Goldstein and Bassi, 2001). Yang et al., (2012) also found that CK17 was totally negative in normal pancreas.
Few studies investigated the expression of Ki-67 in PDAC, with results reflect usage of variable cutoff points for positivity. Kim et al., (2015) found that Ki-67 was positive in 88.2% of the PDAC cases, considering 1 % as a cutoff point for positivity. Whereas a lower figure was reported by Jahng et al., (2010) who considered the intense immunostaining for greater than 50% of population as a cutoff point for positivity. They reported that 41% of PDAC cases were positive for Ki-67, while all benign cases of pancreatitis were negative.
In the present study, the majority of positive PDAC cases for Maspin and CK17 showed diffuse immunostaining and strong intensity. These results come in concordance with many studies (Chu et al., 2005;Cao et al., 2007;Liu et al., 2012;Aksoy-Altinboga et al., 2018). Whereas study conducted by Goldstein and bassi (2001) revealed that focal and diffuse immunostaining for CK17 are equally distributed among positive PDAC cases. Similarly, the majority of our positive PDAC cases for Ki-67 exhibited strong immunostaining intensity, while diffuse immunostaining only represented 22.41%. To the best of our knowledge, no studies in the literature studied the percentage of positive cells and intensity of Ki-67 immunostaining in PDAC.
In our study, for differentiation between PDAC and benign pancreatic tissue, we found that Maspin on its own had the highest sensitivity, NPV and accuracy, when compared with CK17 and Ki-67. Specificity and PPV were 100% for each marker. We demonstrated that combined use of Maspin, CK17 and Ki-67 as a triple panel with at least one of them is positive, achieved the highest sensitivity, NPV and accuracy of 98.8%, 96.2% and 99% for diagnosing PDAC, when compared to using each two of them separately (alternatively double panels). Liu et al., (2012) studied 26 markers in the differentiation between PDAC and normal/non-neoplastic pancreatic tissues. In concordance with our study, all PDAC cases (100%) showed positive staining for Maspin. Whereas CK17 was only positive in 60 % of PDAC cases, which was much lower than we found. Both Maspin and CK17 were negative in all normal/non neoplastic pancreatic tissues. The same findings noticed by Lok et al., (2014) who evaluated immunohistochemical panel in distinction between intrahepatic cholangiocarcinoma and metastatic PDAC on a liver biopsy. CK17 was only seen in 60% of cases of PDAC, while all PDAC cases showed positive Maspin immunostaining.
Furhata et al., (2017) considered a higher cutoff point (26%) for Maspin positivity. They found that Maspin was positive in 40 % of PDAC and Ki-67 was positive in 53 % of PDAC, while no one of normal pancreatic ductal tissues was positive for both markers.
In correlation between our studied markers and studied clinico-pathological data, the only one was found to have statistical significant relation was Ki-67-positive tumors with tumor grade, compared with negative ones. Kim et al., (2015) similarly found that Ki-67 expression was correlated with pathological grade. They also found a correlation with lymph node, metastasis, and clinical stage. Whereas, Karamitopoulou et al., (2010) reported that Ki-67 labeling index did not show any significant association with tumor grade, T stage or lymph node status. In contrast, both Maspin and CK17 immunostaining were not significantly associated with all our studied clinico-pathological data. Cao et al., (2007) also noticed that Maspin expression was not correlated with multiple clinico-pathologic variables. To the best of our knowledge, no studies in the literature correlated the significant relationship between CK17 immunostaining and clinicopathological variables in PDAC.
To conclude, Maspin, CK17 and Ki-67 seem to be helpful markers in increasing the accuracy of diagnosing PDAC in EUS-FNAC. Usage of Maspin, CK17 and Ki-67 as a triple test (at least one of them is positive) is a potential ICC panel in the differentiation between PDAC and benign pancreatic tissue. Out of the 3 studied markers, the only one showed statistically significant relation with tumor grade is Ki-67 positive cases, compared with the Ki-67-negative ones.