Synaptophysin-Ki67 double stain: a novel technique that improves interobserver agreement in the grading of well-differentiated gastrointestinal neuroendocrine tumors

Automated AI-based grading of neuroendocrine tumors using Ki-67 proliferation index: comparative evaluation and performance analysis

Early detection is critical for successfully diagnosing cancer, and timely analysis of diagnostic tests is increasingly important. In the context of neuroendocrine tumors, the Ki-67 proliferation index serves as a fundamental biomarker, aiding pathologists in grading and diagnosing these tumors based on histopathological images. The appropriate treatment plan for the patient is determined based on the tumor grade. An artificial intelligence-based method is proposed to aid pathologists in the automated calculation and grading of the Ki-67 proliferation index. The proposed system first performs preprocessing to enhance image quality. Then, segmentation process is performed using the U-Net architecture, which is a deep learning algorithm, to separate the nuclei from the background. The identified nuclei are then evaluated as Ki-67 positive or negative based on basic color space information and other features. The Ki-67 proliferation index is then calculated, and the neuroendocrine tumor is graded accordingly. The proposed system's performance was evaluated on a dataset obtained from the Department of Pathology at Meram Faculty of Medicine Hospital, Necmettin Erbakan University. The results of the pathologist and the proposed system were compared, and the proposed system was found to have an accuracy of 95% in tumor grading when compared to the pathologist's report.

Ki-67 Proliferation Index Is Associated With Tumor Grade and Survival in Pleural Epithelioid Mesotheliomas

Pleural epithelioid mesothelioma (PEM) is divided into low and high grades based on nuclear atypia, mitoses, and necrosis in the tumor. Assessing mitoses and nuclear atypia tend to be labor-intensive with limited reproducibility. Ki-67 proliferation index was shown to be a prognostic factor in PEM, but its performance has not been directly correlated with tumor grade or mitotic score. This study evaluated the potential of Ki-67 index as a surrogate of tumor grade. We also compared the predictability of mitoses and Ki-67 index for overall survival (OS). Ninety-six PEM samples from 85 patients were identified from the surgical pathology file during 2000-2021 at our institution, and all glass slides were reviewed by 2 pulmonary pathologists to confirm the diagnosis and assign the tumor grade. Digital image analysis (DIA) was done for Ki-67 index. The agreement on tumor grading between 2 reviewers was moderate (kappa value = 0.47). The correlation between mitotic count (average count by 2 reviewers) and Ki-67 index was 0.65. The areas under the curve for predicting tumor grade by mitotic score and Ki-67 index were 0.84 and 0.74 (reviewer 1) and 0.85 and 0.81 (reviewer 2), respectively. High Ki-67 index and mitoses were significantly associated with poor OS ( P =0.03 and 0.0005, using 30% and 10/2 mm 2 as cutoffs, respectively). In conclusion, Ki-67 index by DIA was associated with tumor grade as well as mitotic count, and its predictability for OS was comparable to that of mitotic score, thus being a potential surrogate for tumor grade.

Applications of Deep Learning in Endocrine Neoplasms

Machine learning methods have been growing in prominence across all areas of medicine. In pathology, recent advances in deep learning (DL) have enabled computational analysis of histological samples, aiding in diagnosis and characterization in multiple disease areas. In cancer, and particularly endocrine cancer, DL approaches have been shown to be useful in tasks ranging from tumor grading to gene expression prediction. This review summarizes the current state of DL research in endocrine cancer histopathology with an emphasis on experimental design, significant findings, and key limitations.

Grading of Medullary Thyroid Carcinoma: an Interobserver Reproducibility Study

Grade, based on proliferative activity and tumor necrosis, has recently been shown to be prognostic in medullary thyroid carcinoma (MTC) in multivariate analysis. The aim of this study was to evaluate the interobserver reproducibility of assessed grade in MTC. Three groups (each group included one resident/fellow and one attending pathologist) independently evaluated a cohort of 44 sporadic MTC. For each case, all available tumor slides were reviewed, and mitotic count and the presence of tumor necrosis were recorded. Ki-67 was performed, and the Ki-67 proliferative index was determined in the area of highest proliferative activity. Tumors were graded according to the recently published International Medullary Thyroid Carcinoma Grading System (IMTCGS). Kappa statistics were calculated for each individual criterion (mitotic count, Ki-67 proliferative index, and necrosis) and for assigned IMTCGS grade. For our cohort of 44 MTCs, the kappa statistic for mitotic count, Ki-67 proliferative index, and necrosis was 0.68, 0.86, and 0.89, respectively. The kappa statistic for assigned IMTCGS grade was 0.87. Our findings indicate that there was a strong level of agreement for assessment of grade in our cohort of MTC, indicating that grade as assessed by the IMTCGS is not only prognostic but also reproducible.

Ki-67 assessment of pancreatic neuroendocrine neoplasms: Systematic review and meta-analysis of manual vs. digital pathology scoring

Ki-67 assessment is a key step in the diagnosis of neuroendocrine neoplasms (NENs) from all anatomic locations. Several challenges exist related to quantifying the Ki-67 proliferation index due to lack of method standardization and inter-reader variability. The application of digital pathology coupled with machine learning has been shown to be highly accurate and reproducible for the evaluation of Ki-67 in NENs. We systematically reviewed all published studies on the subject of Ki-67 assessment in pancreatic NENs (PanNENs) employing digital image analysis (DIA). The most common advantages of DIA were improvement in the standardization and reliability of Ki-67 evaluation, as well as its speed and practicality, compared to the current gold standard approach of manual counts from captured images, which is cumbersome and time consuming. The main limitations were attributed to higher costs, lack of widespread availability (as of yet), operator qualification and training issues (if it is not done by pathologists), and most importantly, the drawback of image algorithms counting contaminating non-neoplastic cells and other signals like hemosiderin. However, solutions are rapidly developing for all of these challenging issues. A comparative meta-analysis for DIA versus manual counting shows very high concordance (global coefficient of concordance: 0.94, 95% CI: 0.83-0.98) between these two modalities. These findings support the widespread adoption of validated DIA methods for Ki-67 assessment in PanNENs, provided that measures are in place to ensure counting of only tumor cells either by software modifications or education of non-pathologist operators, as well as selection of standard regions of interest for analysis. NENs, being cellular and monotonous neoplasms, are naturally more amenable to Ki-67 assessment. However, lessons of this review may be applicable to other neoplasms where proliferation activity has become an integral part of theranostic evaluation including breast, brain, and hematolymphoid neoplasms.

Double Ki-67 and synaptophysin labeling in pancreatic neuroendocrine tumor biopsies

BACKGROUND

Pancreatic neuroendocrine tumors (PanNETs) are frequently detected on endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNAB) specimens. The conventional methods for evaluating the Ki-67 labeling index (Ki67LI) in EUS-FNAB specimens are laborious, and their results are difficult to interpret. More practical and easy methods for evaluating the Ki67LI in PanNETs from EUS-FNAB specimens is increasing in need.

METHODS

We used double Ki-67 and synaptophysin (double Ki-Syn) antibody cocktail; Ki67LI, total Ki-67 positive cells, and total tumor cells were counted and compared with those detected on conventional single Ki-67 immunostaining (single Ki-67) of 96 PanNETs [Grade 1 (G1), 68 cases (71%); G2, 26 (27%); G3, 2 (2%)] from EUS-FNAB specimens.

RESULTS

The tumor grading between double Ki-Syn and single Ki-67 immunolabeling was highly concordant (correlation, 0.95; Fisher's exact test, P < 0.001). Seven EUS-FNAB specimens (7%) had discrepant results, of which 2 were removed through surgical resection and showed the same tumor grade as that detected on double Ki-Syn immunolabeling. Fifty-four specimens (56%) had higher Ki-67 positive tumor cell counts on single Ki-67 immunolabeling. Sixty-two specimens (65%) had higher total tumor cell counts on double Ki-Syn immunolabeling. The number of specimens with less than 500 total counted tumor cells were significantly reduced when double Ki-Syn immunolabeling was applied [P = 0.046; single Ki-67, 17 specimens (18%); double Ki-Syn, 9 specimens (9%)].

CONCLUSION

Double Ki-Syn immunolabeling enables the accurate counting of the number of proliferating tumor cells without including inflammatory and contaminant epithelial cells compared with single Ki-67 immunolabeling in PanNETs from EUS-FNAB specimens.

Automated assessment of Ki-67 proliferation index in neuroendocrine tumors by deep learning

The Ki‐67 proliferation index (PI) is a prognostic factor in neuroendocrine tumors (NETs) and defines tumor grade. Analysis of Ki‐67 PI requires calculation of Ki‐67‐positive and Ki‐67‐negative tumor cells, which is highly subjective. To overcome this, we developed a deep learning‐based Ki‐67 PI algorithm (KAI) that objectively calculates Ki‐67 PI. Our study material consisted of NETs divided into training (n = 39), testing (n = 124), and validation (n = 60) series. All slides were digitized and processed in the Aiforia^® Create (Aiforia Technologies, Helsinki, Finland) platform. The ICC between the pathologists and the KAI was 0.89. In 46% of the tumors, the Ki‐67 PIs calculated by the pathologists and the KAI were the same. In 12% of the tumors, the Ki‐67 PI calculated by the KAI was 1% lower and in 42% of the tumors on average 3% higher. The DL‐based Ki‐67 PI algorithm yields results similar to human observers. While the algorithm cannot replace the pathologist, it can assist in the laborious Ki‐67 PI assessment of NETs. In the future, this approach could be useful in, for example, multi‐center clinical trials where objective estimation of Ki‐67 PI is crucial.

Autofluorescence Imaging of Treatment Response in Neuroendocrine Tumor Organoids

Gastroenteropancreatic neuroendocrine tumors (GEP-NET) account for roughly 60% of all neuroendocrine tumors. Low/intermediate grade human GEP-NETs have relatively low proliferation rates that animal models and cell lines fail to recapitulate. Short-term patient-derived cancer organoids (PDCOs) are a 3D model system that holds great promise for recapitulating well-differentiated human GEP-NETs. However, traditional measurements of drug response (i.e., growth, proliferation) are not effective in GEP-NET PDCOs due to the small volume of tissue and low proliferation rates that are characteristic of the disease. Here, we test a label-free, non-destructive optical metabolic imaging (OMI) method to measure drug response in live GEP-NET PDCOs. OMI captures the fluorescence lifetime and intensity of endogenous metabolic cofactors NAD(P)H and FAD. OMI has previously provided accurate predictions of drug response on a single cell level in other cancer types, but this is the first study to apply OMI to GEP-NETs. OMI tested the response to novel drug combination on GEP-NET PDCOs, specifically ABT263 (navitoclax), a Bcl-2 family inhibitor, and everolimus, a standard GEP-NET treatment that inhibits mTOR. Treatment response to ABT263, everolimus, and the combination were tested in GEP-NET PDCO lines derived from seven patients, using two-photon OMI. OMI measured a response to the combination treatment in 5 PDCO lines, at 72 h post-treatment. In one of the non-responsive PDCO lines, heterogeneous response was identified with two distinct subpopulations of cell metabolism. Overall, this work shows that OMI provides single-cell metabolic measurements of drug response in PDCOs to guide drug development for GEP-NET patients.

Improving the accuracy of gastrointestinal neuroendocrine tumor grading with deep learning

The Ki-67 index is an established prognostic factor in gastrointestinal neuroendocrine tumors (GI-NETs) and defines tumor grade. It is currently estimated by microscopically examining tumor tissue single-immunostained (SS) for Ki-67 and counting the number of Ki-67-positive and Ki-67-negative tumor cells within a subjectively picked hot-spot. Intraobserver variability in this procedure as well as difficulty in distinguishing tumor from non-tumor cells can lead to inaccurate Ki-67 indices and possibly incorrect tumor grades. We introduce two computational tools that utilize Ki-67 and synaptophysin double-immunostained (DS) slides to improve the accuracy of Ki-67 index quantitation in GI-NETs: (1) Synaptophysin-KI-Estimator (SKIE), a pipeline automating Ki-67 index quantitation via whole-slide image (WSI) analysis and (2) deep-SKIE, a deep learner-based approach where a Ki-67 index heatmap is generated throughout the tumor. Ki-67 indices for 50 GI-NETs were quantitated using SKIE and compared with DS slide assessments by three pathologists using a microscope and a fourth pathologist via manually ticking off each cell, the latter of which was deemed the gold standard (GS). Compared to the GS, SKIE achieved a grading accuracy of 90% and substantial agreement (linear-weighted Cohen’s kappa 0.62). Using DS WSIs, deep-SKIE displayed a training, validation, and testing accuracy of 98.4%, 90.9%, and 91.0%, respectively, significantly higher than using SS WSIs. Since DS slides are not standard clinical practice, we also integrated a cycle generative adversarial network into our pipeline to transform SS into DS WSIs. The proposed methods can improve accuracy and potentially save a significant amount of time if implemented into clinical practice.

Neuroendocrine Tumors of the Gastrointestinal Tract and Pancreas

Neuroendocrine neoplasms (NENs) of the gastrointestinal (GI) tract and pancreas are a rare and heterogeneous group of neoplasms characterized by common cellular features as well as unique site-specific traits. GI and pancreatic NENs are much rarer than the more common adenocarcinomas arising at these sites. However, the incidences of GI and pancreatic NENs have increased significantly, particularly in the stomach and common site, followed by rectum, appendix, colon, and stomach. Pancreatic NENs are also uncommon, with fewer than 1 per 100,000, accounting for 1% to 2% of all pancreatic neoplasms.

Evaluating the prognostic potential of the Ki67 proliferation index and tumour-infiltrating lymphocytes in olfactory neuroblastoma

AIMS

Olfactory neuroblastomas (ONBs) are rare malignant tumours that arise in the nasal vault. To date, the Hyams grade remains the only widely used histological grading system. However, it is based only on morphological criteria, and has not been updated since 1988. The objective of this study was to explore the prognostic potential of the Ki67 proliferation index (PI) and tumour-infiltrating lymphocytes (TILs) in ONB.

METHODS AND RESULTS

A retrospective study was conducted on a bicentric series of 45 cases. The Ki67 PI was determined by counting at least 1000 nuclei on whole slides. TILs were evaluated with CD20, CD4 and CD8 immunohistochemical markers on whole slides. In this series, Hyams grades I, II, III and IV accounted for 13.4%, 44.4%, 20% and 22.2% of all cases, respectively. The Ki67 PI ranged from 1 to 93; the Ki67 PI was significantly higher in Hyams grade III-IV ONBs than in Hyams grade I-II ONBs (P < 0.0001). A Ki67 PI of ≥25 was associated with poorer survival (P = 0.02). TILs were present in both stromal and intratumoral compartments, but were located predominantly in the stromal component of the tumour. The numbers of intratumoral CD8+ cells/mm² and CD4+ cells/mm² were greater in high-grade ONBs than in low-grade ONBs (P = 0.0015 and P = 0.043, respectively). The numbers of T cells/mm² and B cells/mm² were not associated with survival, but a CD4/CD8 ratio of >2 was significantly associated with shorter survival (P = 0.04).

CONCLUSION

Our findings suggest that the Ki67 PI and TILs could be used as prognostic markers, as a potential alternative to the Hyams grade.

Automated quantification of Ki-67 index associates with pathologic grade of pulmonary neuroendocrine tumors

Background:

Classification of the pulmonary neuroendocrine tumor (pNET) categories is a step-wise process identified by the presence of necrosis and number of mitoses per 2 mm². In neuroendocrine tumor pathology, Ki-67 was first described as a prognostic factor in the pancreas and incorporated into the grading system of digestive tract neuroendocrine neoplasms in the 2010 WHO classification. However, the significance of Ki-67 in pNETs was still a controversial issue. This study was to investigate the potentially diagnostic value of Ki-67 in pNETs.

Methods:

We retrieved 159 surgical specimens of pNETs, including 35 typical carcinoids (TCs), 2 atypical carcinoid (ACs), 28 large-cell neuroendocrine carcinomas (LCNECs), 94 small-cell lung cancers (SCLCs). Manual conventional method (MCM) and computer-assisted image analysis method (CIAM) were used to calculate the Ki-67 proliferative index. In CIAM, 6 equivalent fields (500 × 500 μm) at 10× magnification were manually annotated for digital image analysis.

Results:

The Ki-67 index among the 4 groups with ranges of 0.38% to 12.66% for TC, 4.34% to 29.48% for AC, 30.67% to 93.74% for LCNEC, and 40.71% to 96.87% for SCLC. The cutoff value of Ki-67 index to distinguish low grade with high grade was 30.07%. For the univariate survival analyses in pNETs, both the overall survival and progression-free survival correlated with Ki-67 index. In addition, the Ki-67 index performed by CIAM was proved to be of great positive correlation with MCM.

Conclusions:

Ki-67 index counted by CIAM is a reliable method and can be a useful adjunct to classify the low- and high-grade NETs.

Comparison of Monitor-Image and Printout-Image Methods in Ki-67 Scoring of Gastroenteropancreatic Neuroendocrine Tumors

Gastroenteropancreatic neuroendocrine tumors (GEP-NET) are classified according to tumor grade. Ki-67 and mitotic count are the two determinants of this classification. Therefore, Ki-67 scoring becomes very important in classifying the patients accurately. Eye-balling, counting of cells through the microscope, automated image analysis systems, and manual counting of printed image are the four major scoring methods in use. The aim of this study is to show the agreement between monitor-image method (MIM) and printout-image method (PIM) of Ki-67 scoring. In our study, 120 GEP-NETs from 85 patients diagnosed between January 2005 and July 2017 were evaluated. Thirty-seven cases with either polypectomy or resection material were selected. Seven different scoring methods using either a monitor-image or a printout-image were applied for Ki-67 scoring. They are as follows: whole-PIM, 1/9-PIM, whole-MIM, 1/4-MIM, 1/6-MIM, 1/9-MIM, and 1/12-MIM. In the comparison of Ki-67 scoring methods, intraclass correlation coefficients ranging from 0.951 to 0.999 were found. The Bland-Altman analysis showed near-perfect agreement between whole-MIM and whole-PIM as well as 1/9-MIM and 1/9-PIM. The level of agreements among the other methods were sufficient too, but there was a relative decrease in the level of agreement as the area of counting becomes smaller. The average application time decreased from 373.7 to 41.7 s gradually as the scoring area becomes smaller. Our study shows that there is a remarkable agreement between the MIM and PIM used in Ki-67 scoring.

Immunohistochemical Biomarkers of Gastrointestinal, Pancreatic, Pulmonary, and Thymic Neuroendocrine Neoplasms

Neuroendocrine neoplasms (NENs) are a heterogeneous group of epithelial neoplastic proliferations that irrespective of their primary site share features of neural and endocrine differentiation including the presence of secretory granules, synaptic-like vesicles, and the ability to produce amine and/or peptide hormones. NENs encompass a wide spectrum of neoplasms ranging from well-differentiated indolent tumors to highly aggressive poorly differentiated neuroendocrine carcinomas. Most cases arise in the digestive system and in thoracic organs, i.e., the lung and thymus. A correct diagnostic approach is crucial for the management of patients with both digestive and thoracic NENs, because their high clinical and biological heterogeneity is related to their prognosis and response to therapy. In this context, immunohistochemistry represents an indispensable diagnostic tool that pathologists need to use for the correct diagnosis and classification of such neoplasms. In addition, immunohistochemistry is also useful in identifying prognostic and theranostic markers. In the present article, the authors will review the role of immunohistochemistry in the routine workup of digestive and thoracic NENs.