Gene expression studies have identified molecular subtypes of breast cancer with implications to chemotherapy recommendations. For distinction of these types, a combination of immunohistochemistry (IHC) markers, including proliferative activity of tumor cells, estimated by Ki67 labeling index is used. Clinical studies are frequently based on IHC performed on tissue microarrays (TMA) with variable tissue sampling. This raises the need for evidence-based sampling criteria for individual IHC biomarker studies. We present a novel tissue sampling simulation model and demonstrate its application on Ki67 assessment in breast cancer tissue taking intratumoral heterogeneity into account. Whole slide images (WSI) of 297 breast cancer sections, immunohistochemically stained for Ki67, were subjected to digital image analysis (DIA). Percentage of tumor cells stained for Ki67 was computed for hexagonal tiles super-imposed on the WSI. From this, intratumoral Ki67 heterogeneity indicators (Haralick’s entropy values) were extracted and used to dichotomize the tumors into homogeneous and heterogeneous subsets. Simulations with random selection of hexagons, equivalent to 0.75 mm circular diameter TMA cores, were performed. The tissue sampling requirements were investigated in relation to tumor heterogeneity using linear regression and extended error analysis. The sampling requirements were dependent on the heterogeneity of the biomarker expression. To achieve a coefficient error of 10 %, 5–6 cores were needed for homogeneous cases, 11–12 cores for heterogeneous cases; in mixed tumor population 8 TMA cores were required. Similarly, to achieve the same accuracy, approximately 4,000 nuclei must be counted when the intratumor heterogeneity is mixed/unknown. Tumors of low proliferative activity would require larger sampling (10–12 TMA cores, or 6,250 nuclei) to achieve the same error measurement results as for highly proliferative tumors. Our data show that optimal tissue sampling for IHC biomarker evaluation is dependent on the heterogeneity of the tissue under study and needs to be determined on a per use basis. We propose a method that can be applied to determine the sampling strategy for specific biomarkers, tissues and study targets. In addition, our findings highlight the benefit of high-capacity computer-based IHC measurement techniques to improve accuracy of the testing.

The distribution of tumor-infiltrating lymphocytes (TILs) within the tumor microenvironment provides strong prognostic value, which is increasingly important with the arrival of new immunotherapy modalities. Both visual and image analysis-based assays are developed to assess the immune contexture of the tumors. We propose an automated method based on grid subsampling of microscopy image analysis data to extract the tumor-stroma interface zone (IZ) of controlled width. The IZ is a ranking of tissue areas by their distance to the tumor edge, which is determined by a set of explicit rules. TIL density profiles across the IZ are used to compute a set of novel immunogradient indicators that reflect TIL gradient towards the tumor. We applied the method on CD8 immunohistochemistry images of surgically excised hormone receptor-positive breast and colorectal cancers to predict overall patient survival. In both cohorts, the immunogradient indicators enabled strong and independent prognostic stratification, outperforming clinical and pathologic variables. Patients with breast cancer with low immunogradient levels had a prominent decrease in survival probability 5 years after surgery. Our study provides proof of concept that data-driven, automated, operator-independent IZ sampling enables spatial immune response measurement in the tumor-host interaction frontline for prediction of disease outcomes.

Tumor-associated immune cells have been shown to predict patient outcome in colorectal (CRC) and other cancers. Spatial digital image analysis-based cell quantification increases the informative power delivered by tumor microenvironment features and leads to new prognostic scoring systems. In this study we evaluated the intratumoral density of immunohistochemically stained CD8, CD20 and CD68 cells in 87 cases of CRC (48 were microsatellite stable, MSS, and 39 had microsatellite instability, MSI) in both the intratumoral tumor tissue and within the tumor-stroma interface zone (IZ) which was extracted by a previously developed unbiased hexagonal grid analytics method. Indicators of immune-cell gradients across the extracted IZ were computed and explored along with absolute cell densities, clinicopathological and molecular data, including gene mutation (BRAF, KRAS, PIK3CA) and MSI status. Multiple regression modeling identified (p < 0.0001) three independent prognostic factors: CD8+ and CD20+ Immunogradient indicators, that reflect cell migration towards the tumor, were associated with improved patient survival, while the infiltrative tumor growth pattern was linked to worse patient outcome. These features were combined into CD8-CD20 Immunogradient and immuno-interface scores which outperformed both tumor-node-metastasis (TNM) staging and molecular characteristics, and importantly, revealed high prognostic value both in MSS and MSI CRCs.

Assessment of intratumoral heterogeneity and tumor-host interaction within the tumor microenvironment is becoming increasingly important for innovative cancer therapy decisions because of the unique information it can generate about the state of the disease. However, its assessment and quantification are limited by ambiguous definitions of the tumor-host interface and by human cognitive capacity in current pathology practice. Advances in machine learning and artificial intelligence have opened the field of digital pathology to novel tissue image analytics and feature extraction for generation of high-capacity computational disease management models. A particular benefit is expected from machine-learning applications that can perform extraction and quantification of subvisual features of both intratumoral heterogeneity and tumor microenvironment aspects. These methods generate information about cancer cell subpopulation heterogeneity, potential tumor-host interactions, and tissue microarchitecture, derived from morphologically resolved content using both explicit and implicit features. Several studies have achieved promising diagnostic, prognostic, and predictive artificial intelligence models that often outperform current clinical and pathology criteria. However, further effort is needed for clinical adoption of such methods through development of standardizable high-capacity workflows and proper validation studies.

Background: Bacille Calmette–Guerin (BCG) immunotherapy is the first-line treatment in patients with high-risk non-muscle invasive papillary urothelial carcinoma (NMIPUC), the most common type of bladder cancer. The therapy outcomes are variable and may depend on the immune response within the tumor microenvironment. In our study, we explored the prognostic value of CD8+ cell density gradient indicators across the tumor epithelium–stroma interface of NMIPUC. Methods: Clinical and pathologic data were retrospectively collected from 157 NMIPUC patients treated with BCG immunotherapy after transurethral resection. Whole-slide digital image analysis of CD8 immunohistochemistry slides was used for tissue segmentation, CD8+ cell quantification, and the assessment of CD8+ cell densities within the epithelium–stroma interface. Subsequently, the gradient indicators (center of mass and immunodrop) were computed to represent the density gradient across the interface. Results: By univariable analysis of the clinicopathologic factors, including the history of previous NMIPUC, poor tumor differentiation, and pT1 stage, were associated with shorter RFS (p < 0.05). In CD8+ analyses, only the gradient indicators but not the absolute CD8+ densities were predictive for RFS (p < 0.05). The best-performing cross-validated model included previous episodes of NMIPUC (HR = 4.4492, p = 0.0063), poor differentiation (HR = 2.3672, p = 0.0457), and immunodrop (HR = 5.5072, p = 0.0455). Conclusions: We found that gradient indicators of CD8+ cell densities across the tumor epithelium–stroma interface, along with routine clinical and pathology data, improve the prediction of RFS in NMIPUC.