The importance of digital pathology in the evaluation of proliferation index in pancreas neuroendocrine tumors

Aim: In our study, it was aimed to calculate the Ki67 proliferation activity index, which is necessary for determining the differentiation and grading in neuroendocrine tumors of the pancreas, using three separate counting methods and to determine which method would be more appropriate in routine practice. In addition, it was aimed to show the relationship between pancreatic neuroendocrine tumors and their anatomical localization, histomorphological features and to investigate their effects on prognosis. Materials and methods: Patients who underwent surgery with the preliminary diagnosis of pancreatic neuroendocrine neoplasia between January 2000 and August 2023 and were diagnosed with neuroendocrine tumor or carcinoma as a result of the histopathological examination and immunohistochemical evaluation performed in our department were scanned through the Avicenna program. Accordingly, cases of endocrine tumor originating from the pancreas were evaluated. Hematoxylin-eosin sections, neuroendocrine marker (Synaptophysin and Chromogranin A) and Ki67 stained immunohistochemical sections in our department's archives of cases that met the inclusion criteria were re-evaluated. When evaluating preparations previously stained with hematoxylin and eosin, histomorphological features of pancreatic neuroendocrine neoplasia were described, including the number of mitoses required to confirm the diagnosis. Immunohistochemically stained sections will first be examined under a light microscope, and the Ki67 proliferation index was calculated by counting tumor cells with positive nuclear staining in 2000 tumor cells at x40 magnification in the hot spots where staining is intense in the light microscope. Secondly, counting was done using 3DHistech Pannoramic Scanner (P250 Flash III) with x40 objective. Ki67 immunohistochemistry stained sections were scanned in 3DHistech Pannoramic Scanner (P250 Flash III). The areas with the highest Ki67 immunoreactivity intensity ("hot spots") were manually selected by the pathologist on the scanned slide images at 400x magnification. These areas were exported in high resolution image format using semi-automatic digital analysis software (Case Viewer 2.5 rtm). For all cases, Ki67 immunoreactivity was expressed as the percentage of nuclear-staining tumor cells given by the number of positive tumor cells divided by a minimum of 2000 total tumor cells. Then the automatic program (Quant Center, Cell Quant regime) will evaluate the same hotspot areas selected by the pathologist, and then the result will be given, which will consist of the number of positively stained cells and the total number of cells, expressed as a percentage. A total of 87 cases of pancreatic neuroendocrine neoplasia were evaluated. Patients' age, gender, surgery type, diagnosis, stage, tumor localization, tumor size, presence of tumor in the surgical resection area, presence of necrosis, presence of lymphovascular and/or perineural invasion, lymph node metastasis, and distant metastasis are evaluated. Clinical information such as that of the patients was obtained from the "Hospital Information System" system. To confirm or modify diagnoses, selected hematoxylin-eosin stained sections were redetermined according to the latest 2017 World Health Organization Endocrine classification. Results: The study included 52 female (59.8%) and 35 (40.2%) male patients with an average age of 51.6 years. The age range ranged from 24 to 83 years. 25 G1 PNET cases, 43 G2 PNET cases and 19 G3 PNET and PNEC cases were identified by manual Ki67 counting method with a light microscope. In manual evaluation on digital images, 32 cases were evaluated as G1 PNET, 36 cases as G2 PNET, and 19 cases as G3 PNET and PNEC. In automatic evaluation on digital images, 23 cases were evaluated as G1 PNET, 32 cases as G2 PNET, and 19 cases as G3 PNET and PNEC. When the results obtained regarding the number of mitosis were compared with the Ki67 proliferative index results evaluated under the light microscope, a high correlation was found between the number of mitosis and Ki67 (p<0.0001). In our study, the consistency between the Ki 67 proliferation index results obtained by three different methods for each case: manual evaluation with a light microscope, manual evaluation on a digital image, and automatic evaluation on a digital image was evaluated. Statistically, the intraclass correlation coefficient (ICC) was determined between 0.964 and 0.983 (average 0.975). There appeared to be a high level of correlation between the three methods for calculating the Ki67 proliferation index. The results of the Ki67 proliferative index calculation methods used in our study were found to be statistically significant and compatible with each other (p<0.0001). When automatic evaluation on the digital image and manual evaluation methods on the digital image were used to determine the tumor degree with Ki67 results, the tumor degree was detected to be lower; It was observed that the tumor grade was determined higher when the manual calculation method was used with a light microscope. In comparative analyses, it was determined that there were statistical differences in gender (p<0.05), tumor borders (p<0.001), TNM stage (p<0.001), histological pattern (p<0.001), presence of necrosis (p<0.05), number of mitosis ( p<0.001), presence of additional invasion (p<0.001), PNI (p<0.05), LVI (p<0.001), metastatic LN (p<0.05) and distant metastasis (p<0.001). Age, functional tumor or syndromic tumor presence, tumor localization, positive surgical margin and tumor size were not statistically significant (p>0.05). The average life expectancy in low and moderate PNEN cases was determined to be 160 months after diagnosis. The average survival time in NET grade 3 cases was found to be 18 months after diagnosis. The average life expectancy of patients diagnosed with poorly differentiated NEC was determined to be 25 months after diagnosis. Conclusions: In our study, female predominance in grade 1 and 2 PNETs; It revealed male predominance in grade 3 PNETs and PNEC. In our study, 14 patients (16%) were diagnosed with PNEC and the average age was found to be 54 years. In our study, 79 (95.41%) tumors were sporadic. 4 (4.59%) cases of hereditary syndrome were detected. In our study, 13 (85%) functional tumor cases were identified. In our study, a high correlation was observed when staging results were compared according to Ki67 proliferative index and mitotic count results. In our study, three counting methods (manual counting using a light microscope, manual counting using a digital image and automatic program) were used to evaluate the Ki67 proliferative index, and the ICC average was found to be 0.975. As a result, it was observed that there was a high level of correlation between the three methods for calculating the Ki67 proliferation index. When evaluating the parameters affecting the degree of differentiation of PNEN, 9 parameters were defined: loss of morphological differentiation, character of tumor borders, presence of invasion into neighboring tissues/organs, presence of necrosis, number of mitosis, presence of lymphovascular and perineural invasion, metastasis in lymph nodes and/or distant organs. The patient's age, presence of functional tumor or syndromic tumor, tumor localization, positive surgical margin, and tumor size parameters did not show a relationship with the definition of PNEN differentiation. When evaluating overall survival, the median time after surgical resection revealed that PNET grade 1 and PNET grade 2 showed a high overall survival compared to PNET grade 3 and PNEC grade 3. When the overall survival of PNET grade 3 and PNEC grade 3 was compared, PNET grade 3 showed higher overall survival than PNEC grade 3, despite similar values. The results of our study show that programs based on automatic counting on digital images in determining the Ki67 proliferative index can be used in routine practice for the diagnosis and classification of PNENs. With this; In the evaluation made using all three methods, it should be taken into consideration that counting the positively stained cells to ensure that they are tumor cells may change the result. Keywords: Pancreas, Neuroendocrine Neoplasia, Ki67, Digital analysis