Automated Ki-67 labeling index assessment in prostate cancer using artificial intelligence and multiplex fluorescence immunohistochemistry.

The Ki-67 labeling index (Ki-67 LI) is a strong prognostic marker in prostate cancer, although its analysis requires cumbersome manual quantification of Ki-67 immunostaining in 200-500 tumor cells. To enable automated Ki-67 LI assessment in routine clinical practice, a framework for automated Ki-67 LI quantification, which comprises three different artificial intelligence analysis steps and an algorithm for cell-distance analysis of multiplex fluorescence immunohistochemistry (mfIHC) staining, was developed and validated in a cohort of 12,475 prostate cancers. The prognostic impact of the Ki-67 LI was tested on a tissue microarray (TMA) containing one 0.6 mm sample per patient. A 'heterogeneity TMA' containing three to six samples from different tumor areas in each patient was used to model Ki-67 analysis of multiple different biopsies, and 30 prostate biopsies were analyzed to compare a 'classical' bright field-based Ki-67 analysis with the mfIHC-based framework. The Ki-67 LI provided strong and independent prognostic information in 11,845 analyzed prostate cancers (p < 0.001 each), and excellent agreement was found between the framework for automated Ki-67 LI assessment and the manual quantification in prostate biopsies from routine clinical practice (intraclass correlation coefficient: 0.94 [95% confidence interval: 0.87-0.97]). The analysis of the heterogeneity TMA revealed that the Ki-67 LI of the sample with the highest Gleason score (area under the curve [AUC]: 0.68) was as prognostic as the mean Ki-67 LI of all six foci (AUC: 0.71 [p = 0.24]). The combined analysis of the Ki-67 LI and Gleason score obtained on identical tissue spots showed that the Ki-67 LI added significant additional prognostic information in case of classical International Society of Urological Pathology grades (AUC: 0.82 [p = 0.002]) and quantitative Gleason score (AUC: 0.83 [p = 0.018]). The Ki-67 LI is a powerful prognostic parameter in prostate cancer that is now applicable in routine clinical practice. In the case of multiple cancer-positive biopsies, the sole automated analysis of the worst biopsy was sufficient. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Nonredundant Upregulation of CD112R (PVRIG) and PD-1 on Cytotoxic T Lymphocytes Located in T Cell Nests of Colorectal Cancer.

Focal T lymphocyte aggregates commonly occur in colorectal cancer; however, their biological significance is unknown. To study focal aggregates of T lymphocytes, a deep learning-based framework for automated identification of T cell accumulations (T cell nests) was developed using CD8, PD-1, CD112R, and Ki67 multiplex fluorescence immunohistochemistry. To evaluate the clinical significance of these parameters, a cohort of 523 colorectal cancers with clinical follow-up data was analyzed. Spatial analysis of locally enriched CD8+ T cell density and cell-to-cell contacts identified T cell nests in the tumor microenvironment of colorectal cancer. CD112R and PD-1 expressions on CD8+ T cells located in T cell nests were found to be elevated compared with those on CD8+ T cells in all other tumor compartments (P < .001 each). Although the highest mean CD112R expression on CD8+ T cells was observed at the invasive margin, the PD-1 expression on CD8+ T cells was elevated in the center of the tumor (P < .001 each). Across all tissue compartments, proliferating CD8+ T cells showed higher relative CD112R and PD-1 expressions than those shown by non-proliferating CD8+ T cells (P < .001 each). Integration of all available spatial and immune checkpoint expression parameters revealed a superior predictive performance for overall survival (area under the curve, 0.65; 95% CI, 0.60-0.70) compared with the commonly used CD8+ tumor-infiltrating lymphocyte density (area under the curve, 0.57; 95% CI, 0.53-0.61; P < .001). Cytotoxic T cells with elevated CD112R and PD-1 expression levels are orchestrated in T cell nests of colorectal cancer and predict favorable patient outcomes, and the spatial nonredundancy underlies fundamental differences between both inhibitory immune checkpoints that provide a rationale for dual anti-CD112R/PD-1 immune checkpoint therapy.

Semi-automated validation and quantification of CTLA-4 in 90 different tumor entities using multiple antibodies and artificial intelligence.

CTLA-4 is an inhibitory immune checkpoint receptor and a negative regulator of anti-tumor T-cell function. This study is aimed for a comparative analysis of CTLA-4+ cells between different tumor entities. To quantify CTLA-4+ cells, 4582 tumor samples from 90 different tumor entities as well as 608 samples of 76 different normal tissue types were analyzed by immunohistochemistry in a tissue microarray format. Two different antibody clones (MSVA-152R and CAL49) were validated and quantified using a deep learning framework for automated exclusion of unspecific immunostaining. Comparing both CTLA-4 antibodies revealed a clone dependent unspecific staining pattern in adrenal cortical adenoma (63%) for MSVA-152R and in pheochromocytoma (67%) as well as hepatocellular carcinoma (36%) for CAL49. After automated exclusion of non-specific staining reaction (3.6%), a strong correlation was observed for the densities of CTLA-4+ lymphocytes obtained by both antibodies (r = 0.87; p < 0.0001). A high CTLA-4+ cell density was linked to low pT category (p < 0.0001), absent lymph node metastases (p = 0.0354), and PD-L1 expression in tumor cells or inflammatory cells (p < 0.0001 each). A high CTLA-4/CD3-ratio was linked to absent lymph node metastases (p = 0.0295) and to PD-L1 positivity on immune cells (p = 0.0026). Marked differences exist in the number of CTLA-4+ lymphocytes between tumors. Analyzing two independent antibodies by a deep learning framework can facilitate automated quantification of immunohistochemically analyzed target proteins such as CTLA-4.

Prognostic Impact and Spatial Interplay of Immune Cells in Urothelial Cancer.

BACKGROUND AND OBJECTIVE: Quantity and the spatial relationship of specific immune cell types can provide prognostic information in bladder cancer. The objective of the study was to characterize the spatial interplay and prognostic role of different immune cell subpopulations in bladder cancer. METHODS: A total of 2463 urothelial bladder carcinomas were immunostained with 21 antibodies using BLEACH&STAIN multiplex fluorescence immunohistochemistry in a tissue microarray format and analyzed using a framework of neuronal networks for an image analysis. Spatial immune parameters were compared with histopathological parameters and overall survival data. KEY FINDINGS AND LIMITATIONS: The identification of > 300 different immune cell subpopulations and the characterization of their spatial relationship resulted in numerous spatial interaction patterns. Thirty-nine immune parameters showed prognostic significance in univariate analyses, of which 16 were independent from pT, pN, and histological grade in muscle-invasive bladder cancer. Among all these parameters, the strongest association with prolonged overall survival was identified for intraepithelial CD8+ cytotoxic T cells (time-dependent area under receiver operating characteristic curve [AUC]: 0.70), while stromal CD8+ T cells were less relevant (AUC: 0.65). A favorable prognosis of inflamed cancers with high levels of "exhaustion markers" suggests that TIM3, PD-L1, PD-1, and CTLA-4 on immune cells do not hinder antitumoral immune response in tumors rich of tumor infiltrating immune cells. CONCLUSIONS AND CLINICAL IMPLICATIONS: The density of intraepithelial CD8+ T cells was the strongest prognostic feature in muscle-invasive bladder cancer. Given that tumor cell killing by CD8+ cytotoxic T lymphocytes through direct cell-to-cell-contacts represents the "terminal end route" of antitumor immunity, the quantity of "tumor cell adjacent CD8+ T cells" may constitute a surrogate for the efficiency of cancer recognition by the immune system that can be measured straightaway in routine pathology as the CD8 labeling index. PATIENT SUMMARY: Quantification of intraepithelial CD8+ T cells, the strongest prognosticfeature identified in muscle-invasive bladder cancer, can easily be assessed by brightfield immunohistochemistry and is therefore "ready to use" for routine pathology.

BLEACH&STAIN 15-marker Multiplexed Imaging in 3,098 Human Carcinomas Reveals Six Major PD-L1-driven Immune Phenotypes with Distinct Spatial Orchestration.

Multiplex fluorescence IHC (mfIHC) approaches were yet either limited to six markers or limited to a small tissue size that hampers translational studies on large tissue microarray cohorts. Here we have developed a BLEACH&STAIN mfIHC method that enabled the simultaneous analysis of 15 biomarkers (PD-L1, PD-1, CTLA-4, panCK, CD68, CD163, CD11c, iNOS, CD3, CD8, CD4, FOXP3, CD20, Ki67, and CD31) in 3,098 tumor samples from 44 different carcinoma entities within one week. To facilitate automated immune checkpoint quantification on tumor and immune cells and study its spatial interplay an artificial intelligence-based framework incorporating 17 different deep-learning systems was established. Unsupervised clustering showed that the three PD-L1 phenotypes (PD-L1+ tumor and immune cells, PD-L1+ immune cells, PD-L1-) were either inflamed or noninflamed. In inflamed PD-L1+patients, spatial analysis revealed that an elevated level of intratumoral M2 macrophages as well as CD11c+ dendritic cell (DC) infiltration (P < 0.001 each) was associated with a high CD3+ CD4± CD8± FOXP3± T-cell exclusion and a high PD-1 expression on T cells (P < 0.001 each). In breast cancer, the PD-L1 fluorescence intensity on tumor cells showed a significantly higher predictive performance for overall survival (OS; AUC, 0.72, P < 0.001) compared with the commonly used percentage of PD-L1+ tumor cells (AUC, 0.54). In conclusion, our deep-learning-based BLEACH&STAIN framework facilitates rapid and comprehensive assessment of more than 60 spatially orchestrated immune cell subpopulations and its prognostic relevance. IMPLICATIONS: The development of an easy-to-use high-throughput 15+1 multiplex fluorescence approach facilitates the in-depth understanding of the immune tumor microenvironment (TME) and enables to study the prognostic relevance of more than 130 immune cell subpopulations.

Automated Prognosis Marker Assessment in Breast Cancers Using BLEACH&STAIN Multiplexed Immunohistochemistry.

Prognostic markers in routine clinical management of breast cancer are often assessed using RNA-based multi-gene panels that depend on fluctuating tumor purity. Multiplex fluorescence immunohistochemistry (mfIHC) holds the potential for an improved risk assessment. To enable automated prognosis marker detection (i.e., progesterone receptor [PR], estrogen receptor [ER], androgen receptor [AR], GATA3, TROP2, HER2, PD-L1, Ki67, TOP2A), a framework for automated breast cancer identification was developed and validated involving thirteen different artificial intelligence analysis steps and an algorithm for cell distance analysis using 11+1-marker-BLEACH&STAIN-mfIHC staining in 1404 invasive breast cancers of no special type (NST). The framework for automated breast cancer detection discriminated normal glands from malignant glands with an accuracy of 98.4%. This approach identified that five (PR, ER, AR, GATA3, PD-L1) of nine biomarkers were associated with prolonged overall survival (p ≤ 0.0095 each) and two of these (PR, AR) were found to be independent risk factors in multivariate analysis (p ≤ 0.0151 each). The combined assessment of PR-ER-AR-GATA3-PD-L1 as a five-marker prognosis score showed strong prognostic relevance (p < 0.0001) and was an independent risk factor in multivariate analysis (p = 0.0034). Automated breast cancer detection in combination with an artificial intelligence-based analysis of mfIHC enables a rapid and reliable analysis of multiple prognostic parameters. The strict limitation of the analysis to malignant cells excludes the impact of fluctuating tumor purity on assay precision.

Loss of TROP2 and epithelial cell adhesion molecule expression is linked to grade progression in pTa but unrelated to disease outcome in pT2-4 urothelial bladder carcinomas.

Introduction: Trophoblast cell surface antigen 2 (TROP2; EpCAM2) is a transmembrane glycoprotein which is closely related to EpCAM (EpCAM; EpCAM1). Both proteins share partial overlapping functions in epithelial development and EpCAM expression but have not been comparatively analyzed together in bladder carcinomas. TROP2 constitutes the target for the antibody-drug conjugate Sacituzumab govitecan (SG; TrodelvyTM) which has been approved for treatment of metastatic urothelial carcinoma by the United States Food and Drug administration (FDA) irrespective of its TROP2 expression status. Methods: To evaluate the potential clinical significance of subtle differences in TROP2 and EpCAM expression in urothelial bladder cancer, both proteins were analyzed by multiplex fluorescence immunohistochemistry in combination with a deep-learning based algorithm for automated cell detection on more than 2,700 urothelial bladder carcinomas in a tissue microarray (TMA) format. Results: The staining pattern of TROP2 and EpCAM were highly similar. For both proteins, the staining intensity gradually decreased from pTa G2 low grade (TROP2: 68.8±36.1; EpCAM: 21.5±11.7) to pTa G2 high grade (64.6±38.0; 19.3±12.2) and pTa G3 (52.1±38.7; 16.0±13.0, p<0.001 each). In pT2-4 carcinomas, the average TROP2 and EpCAM staining intensity was intermediate (61.8±40.9; 18.3±12.3). For both proteins, this was significantly lower than in pTa G2 low grade (p<0.001 each) but also higher than in pTa G3 tumors (p=0.022 for TROP2, p=0.071 for EpCAM). Within pT2-4 carcinomas, the TROP2 and EpCAM staining level was unrelated to pT, grade, UICC-category, and overall or tumor-specific patient survival. The ratio TROP2/EpCAM was unrelated to malignant phenotype and patient prognosis. Conclusion: Our data show that TROP2 and EpCAM expression is common and highly interrelated in urothelial neoplasms. Despite of a progressive loss of TROP2/EpCAM during tumor cell dedifferentiation in pTa tumors, the lack of associations with clinicopathological parameters in pT2-4 cancer argues against a major cancer driving role of both proteins for the progression of urothelial neoplasms.

Prognostic role of proliferating CD8+ cytotoxic Tcells in human cancers.

PURPOSE: Expansion of CD8+ cytotoxic Tlymphocytes is a prerequisite for anti-cancer immune activity and has gained interest in the era of immune checkpoint therapy. METHODS: To understand the CD8+ T cell dynamics in the tumor microenvironment, we used multiplex fluorescence immunohistochemistry to quantitate CD8+ proliferation (Ki67 co-expression) in tissue microarrays from 1107 colorectal, 642 renal cell, 1066 breast, 375 ovarian, 451 pancreatic and 347 gastric cancer samples. RESULTS: The density and the percentage of proliferating (Ki67+) CD8+ T cells were both highly variable between tumor types as well as between patients with the same tumor type. Elevated density and percentage of proliferating CD8+ cytotoxic T cells were significantly associated with favorable tumor parameters such as low tumor stage, negative nodal stage (p ≤ 0.0041 each), prolonged overall survival (p ≤ 0.0028 each) and an inflamed immune phenotype (p = 0.0025) in colorectal cancer and, in contrast, linked to high tumor stage, advanced ISUP/Fuhrman/Thoenes grading (each p ≤ 0.003), shorter overall survival (p ≤ 0.0330 each) and an immune inflamed phenotype (p = 0.0094) in renal cell cancer. In breast, ovarian, pancreatic and gastric cancer the role of (Ki67+)CD8+ Tcells was not linked to clinicopathological data. CONCLUSION: Our data demonstrate a tumor type dependent prognostic impact of proliferating (Ki67+)CD8+ Tcells and an inverse impact in colorectal and renal cell cancer.