Machine-Learning-Based Evaluation of Intratumoral Heterogeneity and Tumor-Stroma Interface for Clinical Guidance

Polychromatic Polarization Microscopy Differentiates Collagen Fiber Signatures in Benign Pancreatic Tissue and Pancreatic Ductal Adenocarcinoma

The orientation of collagen fibers in relation to malignant epithelium is known to carry prognostic information in a variety of tissues. The data are the strongest for breast and pancreatic ductal adenocarcinoma. However, information inherent in collagen fiber topology in malignant tissues remains untapped in daily surgical pathology practice, largely because collagen fibers within areas of desmoplasia cannot be resolved with standard diagnostic microscopy. The methodologies used to visualize collagen fiber orientation are either of insufficient resolution to consistently capture collagen fiber topology or require resources in time and money that do not fit into the daily surgical pathology workflow. Polychromatic polarization microscopy has the potential to bring collagen topology to the attention of pathologists during their routine work. It has been demonstrated to be equivalent to the gold standard methodology used to research collagen, second harmonic generation. We use polychromatic polarization microscopy to visualize and describe the differences in collagen topology in normal pancreas, chronic pancreatitis, and pancreatic ductal adenocarcinoma with a standard microscope, using hematoxylin and eosin-stained sections. In the process, we propose a lexicon with which to describe the morphologic characteristics of collagen in benign and malignant pancreatic tissues.

Critical review of patient outcome study in head and neck cancer radiotherapy

Rapid technological advances in radiation therapy have significantly improved dose delivery and tumor control for head and neck cancers. However, treatment-related toxicities caused by high-dose exposure to critical structures remain a significant clinical challenge, underscoring the need for accurate prediction of clinical outcomes-encompassing both tumor control and adverse events (AEs). This review critically evaluates the evolution of data-driven approaches in predicting patient outcomes in head and neck cancer patients treated with radiation therapy, from traditional dose-volume constraints to cutting-edge artificial intelligence (AI) and causal inference framework. The integration of linear energy transfer in patient outcomes study, which has uncovered critical mechanisms behind unexpected toxicity, was also introduced for proton therapy. Three transformative methodological advances are reviewed: radiomics, AI-based algorithms, and causal inference frameworks. While radiomics has enabled quantitative characterization of medical images, AI models have demonstrated superior capability than traditional models. However, the field faces significant challenges in translating statistical correlations from real-world data into interventional clinical insights. We highlight that how causal inference methods can bridge this gap by providing a rigorous framework for identifying treatment effects. Looking ahead, we envision that combining these complementary approaches, especially the interventional prediction models, will enable more personalized treatment strategies, ultimately improving both tumor control and quality of life for head and neck cancer patients treated with radiation therapy.

Characterization of Breast Cancer Intra-Tumor Heterogeneity Using Artificial Intelligence

Intra-tumor heterogeneity (ITH) is a fundamental characteristic of breast cancer (BC), influencing tumor progression, prognosis, and therapeutic responses. However, the complexity of ITH in BC makes its accurate characterization challenging. This study leverages deep learning (DL) techniques to comprehensively evaluate ITH in early-stage luminal BC and provide a nuanced understanding of its impact on tumor behavior and patient outcomes. A large cohort (n = 2561) of early-stage luminal BC was evaluated using whole slide images (WSIs) of hematoxylin and eosin-stained slides of excision specimens. Morphological features of both the tumor and stromal components were meticulously annotated by a panel of pathologists in a subset of cases. A DL model was applied to develop an algorithm to assess the degree of heterogeneity of various morphological features per individual case utilizing defined patches. The results of extracted features were used to generate an overall heterogeneity score that was correlated with the clinicopathological features and outcome. Overall, 162 features were quantified and a significant positive correlation between these features was identified. Specifically, there was a significant association between a high degree of intra-tumor heterogeneity and larger tumor size, poorly differentiated tumors, highly proliferative tumors, tumors of no special type (NST), and those with low estrogen receptor (ER) expression. When all features are considered in combination, a high overall heterogeneity score was significantly associated with parameters characteristic of aggressive tumor behavior, and it was an independent predictor of poor patient outcome. In conclusion, DL models can be used to accurately decipher the complexity of ITH and provide extra information for outcome prediction.

Tumor-stroma contact ratio - a novel predictive factor for tumor response to chemoradiotherapy in locally advanced oropharyngeal cancer

The growth pattern of oropharyngeal squamous cell carcinomas (OPSCC) varies from compact tumor cell aggregates to diffusely infiltrating tumor cell-clusters. The influence of the growth pattern on local tumor control and survival has been studied mainly for surgically treated oral cavity carcinomas on a visual basis. In this study, we used multiplex immunofluorescence staining (mIF) to examine the antigens pan-cytokeratin, p16INK4a, Ki67, CD271, PD-L1, and CD8 in pretherapeutic biopsies from 86 OPSCC. We introduce Tumor-stroma contact ratio (TSC), a novel parameter, to quantify the relationship between tumor cells in contact with the stromal surface and the total number of epithelial tumor cells. mIF tumor cores were analyzed at the single-cell level, and tumor-stromal contact area was quantified using the R package "Spatstat". TSC was correlated with the visually assessed invasion pattern by two independent investigators. Furthermore, TSC was analyzed in relation to clinical parameters and patient survival data to evaluate its potential prognostic significance. Higher TSC correlated with poor response to (chemo-)radiotherapy (r = 0.3, p < 0.01), and shorter overall (OS) and progression-free (PFS) survival (median OS: 13 vs 136 months, p < 0.0001; median PFS: 5 vs 85 months, p < 0.0001). Visual categorization of growth pattern according to established criteria of tumor aggressiveness showed interobserver variability increasing with more nuanced categories (2 categories: k = 0.7, 95 %-CI: 0.55 - 0.85; 4 categories k = 0.48, 95 %-CI: 0.35 - 0.61). In conclusion, TSC is an objective and reproducible computer-based parameter to quantify tumor-stroma contact area. We demonstrate its relevance for the response of oropharyngeal carcinomas to primary (chemo-)radiotherapy.

The classification of benign and malignant lung nodules based on CT radiomics: a systematic review, quality score assessment, and meta-analysis

Radiomics methods are increasingly used to identify benign and malignant lung nodules, and early monitoring is essential in prognosis and treatment strategy formulation. To evaluate the diagnostic performance of computed tomography (CT)-based radiomics for distinguishing between benign and malignant lung nodules by performing a meta-analysis. Between January 2000 and December 2021, we searched the PubMed and Embase electronic databases for studies in English. Studies were included if they demonstrated the sensitivity and specificity of CT-based radiomics for diagnosing benign and malignant lung nodules. The studies were evaluated using the QUADAS-2 and radiomics quality scores (RQS). The inhomogeneity of the data and publishing bias were also evaluated. Some subgroup analyses were performed to investigate the impact of diagnostic efficiency. The Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) Guidelines were followed for this meta-analysis. A total of 20 studies involving 3793 patients were included. The combined sensitivity, specificity, diagnostic odds ratio, and area under the summary receiver operating characteristic curve based on CT radiomics diagnosis of benign and malignant lung nodules were 0.81, 0.86, 27.00, and 0.91, respectively. Deek's funnel plot asymmetry test confirmed no significant publication bias in all studies. Fagan nomograms showed a 40% increase in post-test probability among pretest-positive patients. Current evidence shows that CT-based radiomics has high accuracy in the diagnosis of benign and malignant lung nodules.

Standardizing analysis of intra-tumoral heterogeneity with computational pathology

Many malignant cancers like glioblastoma are highly adaptive diseases that dynamically change their regional biology to survive and thrive under diverse microenvironmental and therapeutic pressures. While the concept of intra-tumoral heterogeneity has become a major paradigm in cancer research and care, systematic approaches to assess and document bio-variation in cancer are still in their infancy. Here we discuss existing approaches and challenges to documenting intra-tumoral heterogeneity and emerging computational approaches that leverage artificial intelligence to begin to overcome these limitations. We propose how these emerging techniques can be coupled with a diversity of molecular tools to address intra-tumoral heterogeneity more systematically in research and in practice, especially across larger specimens and longitudinal analyses. Systematic documentation and characterization of heterogeneity across entire tumor specimens and their longitudinal evolution has the potential to improve our understanding and treatment of cancer.

Machine Learning Improves the Prediction of Responses to Immune Checkpoint Inhibitors in Metastatic Melanoma

Pretreatment LDH is a standard prognostic biomarker for advanced melanoma and is associated with response to ICI. We assessed the role of machine learning-based radiomics in predicting responses to ICI and in complementing LDH for prognostication of metastatic melanoma. From 2008-2022, 79 patients with 168 metastatic hepatic lesions were identified. All patients had arterial phase CT images 1-month prior to initiation of ICI. Response to ICI was assessed on follow-up CT at 3 months using RECIST criteria. A machine learning algorithm was developed using radiomics. Maximum relevance minimum redundancy (mRMR) was used to select features. ROC analysis and logistic regression analyses evaluated performance. Shapley additive explanations were used to identify the variables that are the most important in predicting a response. mRMR selection revealed 15 features that are associated with a response to ICI. The machine learning model combining both radiomics features and pretreatment LDH resulted in better performance for response prediction compared to models that included radiomics or LDH alone (AUC of 0.89 (95% CI: [0.76-0.99]) vs. 0.81 (95% CI: [0.65-0.94]) and 0.81 (95% CI: [0.72-0.91]), respectively). Using SHAP analysis, LDH and two GLSZM were the most predictive of the outcome. Pre-treatment CT radiomic features performed equally well to serum LDH in predicting treatment response.

Prognostic Value of CD8+ Lymphocytes in Hepatocellular Carcinoma and Perineoplastic Parenchyma Assessed by Interface Density Profiles in Liver Resection Samples

Hepatocellular carcinoma (HCC) often emerges in the setting of long-standing inflammatory liver disease. CD8 lymphocytes are involved in both the antitumoral response and hepatocyte damage in the remaining parenchyma. We investigated the dual role of CD8 lymphocytes by assessing density profiles at the interfaces of both HCC and perineoplastic liver parenchyma with surrounding stroma in whole-slide immunohistochemistry images of surgical resection samples. We applied a hexagonal grid-based digital image analysis method to sample the interface zones and compute the CD8 density profiles within them. The prognostic value of the indicators was explored in the context of clinicopathological, peripheral blood testing, and surgery data. Independent predictors of worse OS were a low standard deviation of CD8+ density along the tumor edge, high mean CD8+ density within the epithelial aspect of the perineoplastic liver-stroma interface, longer duration of surgery, a higher level of aspartate transaminase (AST), and a higher basophil count in the peripheral blood. A combined score, derived from these five independent predictors, enabled risk stratification of the patients into three prognostic categories with a 5-year OS probability of 76%, 40%, and 8%. Independent predictors of longer RFS were stage pT1, shorter duration of surgery, larger tumor size, wider tumor-free margin, and higher mean CD8+ density in the epithelial aspect of the tumor-stroma interface. We conclude that (1) our computational models reveal independent and opposite prognostic impacts of CD8+ cell densities at the interfaces of the malignant and non-malignant epithelium interfaces with the surrounding stroma; and (2) together with pathology, surgery, and laboratory data, comprehensive prognostic models can be constructed to predict patient outcomes after liver resection due to HCC.

Deciphering Tumour Heterogeneity: From Tissue to Liquid Biopsy

Simple Summary

Most malignant tumours are highly heterogeneous at molecular and phenotypic levels. Tumour variability poses challenges for the management of patients, as it arises between patients and even evolves in space and time within a single patient. Currently, treatment-decision making usually relies on the molecular characteristics of a limited tumour tissue sample at the time of diagnosis or disease progression but does not take into account the complexity of the bulk tumours and their constant evolution over time. In this review, we explore the extent of tumour heterogeneity and report the mechanisms that promote and sustain this diversity in cancers. We summarise the clinical strikes of tumour diversity in the management of patients with cancer. Finally, we discuss the current material and technological approaches that are relevant to adequately appreciate tumour heterogeneity.

Abstract

Human solid malignancies harbour a heterogeneous set of cells with distinct genotypes and phenotypes. This heterogeneity is installed at multiple levels. A biological diversity is commonly observed between tumours from different patients (inter-tumour heterogeneity) and cannot be fully captured by the current consensus molecular classifications for specific cancers. To extend the complexity in cancer, there are substantial differences from cell to cell within an individual tumour (intra-tumour heterogeneity, ITH) and the features of cancer cells evolve in space and time. Currently, treatment-decision making usually relies on the molecular characteristics of a limited tumour tissue sample at the time of diagnosis or disease progression but does not take into account the complexity of the bulk tumours and their constant evolution over time. In this review, we explore the extent of tumour heterogeneity with an emphasis on ITH and report the mechanisms that promote and sustain this diversity in cancers. We summarise the clinical strikes of ITH in the management of patients with cancer. Finally, we discuss the current material and technological approaches that are relevant to adequately appreciate ITH.

L’intelligence artificielle pourrait-elle intervenir dans l’aide au diagnostic des cancers du sein ? – L’exemple de HER2

HER2 is an important prognostic and predictive biomarker in breast cancer. Its detection makes it possible to define which patients will benefit from a targeted treatment. While assessment of HER2 status by immunohistochemistry in positive vs negative categories is well implemented and reproducible, the introduction of a new "HER2-low" category could raise some concerns about its scoring and reproducibility. We herein described the current HER2 testing methods and the application of innovative machine learning techniques to improve these determinations, as well as the main challenges and opportunities related to the implementation of digital pathology in the up-and-coming AI era.

Intratumoral Heterogeneity and Immune Response Indicators to Predict Overall Survival in a Retrospective Study of HER2-Borderline (IHC 2+) Breast Cancer Patients

Breast cancer (BC) categorized as human epidermal growth factor receptor 2 (HER2) borderline [2+ by immunohistochemistry (IHC 2+)] presents challenges for the testing, frequently obscured by intratumoral heterogeneity (ITH). This leads to difficulties in therapy decisions. We aimed to establish prognostic models of overall survival (OS) of these patients, which take into account spatial aspects of ITH and tumor microenvironment by using hexagonal tiling analytics of digital image analysis (DIA). In particular, we assessed the prognostic value of Immunogradient indicators at the tumor–stroma interface zone (IZ) as a feature of antitumor immune response. Surgical excision samples stained for estrogen receptor (ER), progesterone receptor (PR), Ki67, HER2, and CD8 from 275 patients with HER2 IHC 2+ invasive ductal BC were used in the study. DIA outputs were subsampled by HexT for ITH quantification and tumor microenvironment extraction for Immunogradient indicators. Multiple Cox regression revealed HER2 membrane completeness (HER2 MC) (HR: 0.18, p = 0.0007), its spatial entropy (HR: 0.37, p = 0.0341), and ER contrast (HR: 0.21, p = 0.0449) as independent predictors of better OS, with worse OS predicted by pT status (HR: 6.04, p = 0.0014) in the HER2 non-amplified patients. In the HER2-amplified patients, HER2 MC contrast (HR: 0.35, p = 0.0367) and CEP17 copy number (HR: 0.19, p = 0.0035) were independent predictors of better OS along with worse OS predicted by pN status (HR: 4.75, p = 0.0018). In the non-amplified tumors, three Immunogradient indicators provided the independent prognostic value: CD8 density in the tumor aspect of the IZ and CD8 center of mass were associated with better OS (HR: 0.23, p = 0.0079 and 0.14, p = 0.0014, respectively), and CD8 density variance along the tumor edge predicted worse OS (HR: 9.45, p = 0.0002). Combining these three computational indicators of the CD8 cell spatial distribution within the tumor microenvironment augmented prognostic stratification of the patients. In the HER2-amplified group, CD8 cell density in the tumor aspect of the IZ was the only independent immune response feature to predict better OS (HR: 0.22, p = 0.0047). In conclusion, we present novel prognostic models, based on computational ITH and Immunogradient indicators of the IHC biomarkers, in HER2 IHC 2+ BC patients.