Quantitative histology analysis of the ovarian tumour microenvironment

The endothelin-1-driven tumor-stroma feed-forward loops in high-grade serous ovarian cancer

The high-grade serous ovarian cancer (HG-SOC) tumor microenvironment (TME) is constellated by cellular elements and a network of soluble constituents that contribute to tumor progression. In the multitude of the secreted molecules, the endothelin-1 (ET-1) has emerged to be implicated in the tumor/TME interplay; however, the molecular mechanisms induced by the ET-1-driven feed-forward loops (FFL) and associated with the HG-SOC metastatic potential need to be further investigated. The tracking of the patient-derived (PD) HG-SOC cell transcriptome by RNA-seq identified the vascular endothelial growth factor (VEGF) gene and its associated signature among those mostly up-regulated by ET-1 and down-modulated by the dual ET-1R antagonist macitentan. Within the ligand-receptor pairs concurrently expressed in PD-HG-SOC cells, endothelial cells and activated fibroblasts, we discovered two intertwined FFL, the ET-1/ET-1R and VEGF/VEGF receptors, concurrently activated by ET-1 and shutting-down by macitentan, or by the anti-VEGF antibody bevacizumab. In parallel, we observed that ET-1 fine-tuned the tumoral and stromal secretome toward a pro-invasive pattern. Into the fray of the HG-SOC/TME double and triple co-cultures, the secretion of ET-1 and VEGF, that share a common co-regulation, was inhibited upon the administration of macitentan. Functionally, macitentan, mimicking the effect of bevacizumab, interfered with the HG-SOC/TME FFL-driven communication that fuels the HG-SOC invasive behavior. The identification of ET-1 and VEGF FFL as tumor and TME actionable vulnerabilities, reveals how ET-1R blockade, targeting the HG-SOC cells and the TME simultaneously, may represent an effective therapeutic option for HG-SOC patients.

Spatiotemporal architecture of immune cells and cancer-associated fibroblasts in high-grade serous ovarian carcinoma

High-grade serous ovarian carcinoma (HGSOC), the deadliest form of ovarian cancer, is typically diagnosed after it has metastasized and often relapses after standard-of-care platinum-based chemotherapy, likely due to advanced tumor stage, heterogeneity, and immune evasion and tumor-promoting signaling from the tumor microenvironment. To understand how spatial heterogeneity contributes to HGSOC progression and early relapse, we profiled an HGSOC tissue microarray of patient-matched longitudinal samples from 42 patients. We found spatial patterns associated with early relapse, including changes in T cell localization, malformed tertiary lymphoid structure (TLS)-like aggregates, and increased podoplanin-positive cancer-associated fibroblasts (CAFs). Using spatial features to compartmentalize the tissue, we found that plasma cells distribute in two different compartments associated with TLS-like aggregates and CAFs, and these distinct microenvironments may account for the conflicting reports about the role of plasma cells in HGSOC prognosis.

The prognostic value of tumor-stroma ratio and a newly developed computer-aided quantitative analysis of routine H&E slides in high-grade serous ovarian cancer

INTRODUCTION

Tumor-stroma ratio (TSR) is prognostic in multiple cancers, while its role in high-grade serous ovarian cancer (HGSOC) remains unclear. Despite the prognostic insight gained from genetic profiles and tumor-infiltrating lymphocytes (TILs), the prognostic use of histology slides remains limited, while it enables the identification of tumor characteristics via computational pathology reducing scoring time and costs. To address this, this study aimed to assess TSR's prognostic role in HGSOC and its association with TILs. We additionally developed an algorithm, Ovarian-TSR (OTSR), using deep learning for TSR scoring, comparing it to manual scoring. Methods : 340 patients with advanced-stage who underwent primary debulking surgery (PDS) or neo-adjuvant chemotherapy (NACT) with interval debulking (IDS). TSR was assessed in both the most invasive (MI) and whole tumor (WT) regions through manual scoring by pathologists and quantification using OTSR. Patients were categorized as stroma-rich (≥ 50% stroma) or stroma-poor (< 50%). TILs were evaluated via immunohistochemical staining. Result s: In PDS, stroma-rich tumors were significantly associated with a more frequent papillary growth pattern (60% vs 34%), while In NACT stroma-rich tumors had a lower Tumor Regression Grading (TRG 4&5, 21% vs 57%) and increased pleural metastasis (25% vs 16%). Stroma-rich patients had significantly shorter overall and progression-free survival compared to stroma-poor (31 versus 45 months; P < 0.0001, and 15 versus 17 months; P = 0.0008, respectively). Combining stromal percentage and TILs led to three distinct survival groups with good (stroma-poor, high TIL), medium (stroma-rich, high TIL, or; stroma-poor, Low TIL), and poor(stroma-rich, low TIL) survival. These survival groups remained significant in CD8 and CD103 in multivariable analysis (Hazard ratio (HR) = 1.42, 95% Confidence-interval (CI) = 1.02-1.99; HR = 1.49, 95% CI = 1.01-2.18, and HR = 1.48, 95% CI = 1.05-2.08; HR = 2.24, 95% CI = 1.55-3.23, respectively). OTSR was able to recapitulate these results and demonstrated high concordance with expert pathologists (correlation = 0.83). Conclusions : TSR is an independent prognostic factor for survival assessment in HGSOC. Stroma-rich tumors have a worse prognosis and, in the case of NACT, a higher likelihood of pleural metastasis. OTSR provides a cost and time-efficient way of determining TSR with high reproducibility and reduced inter-observer variability.

Profiling ovarian cancer tumor and microenvironment during disease progression for cell-based immunotherapy design

Ovarian cancer (OC) is highly lethal due to late detection and frequent recurrence. Initial treatments, comprising surgery and chemotherapy, lead to disease remission but are invariably associated with subsequent relapse. The identification of novel therapies and an improved understanding of the molecular and cellular characteristics of OC are urgently needed. Here, we conducted a comprehensive analysis of primary tumor cells and their microenvironment from 16 chemonaive and 10 recurrent OC patient samples. Profiling OC tumor biomarkers allowed for the identification of potential molecular targets for developing immunotherapies, while profiling the microenvironment yielded insights into its cellular composition and property changes between chemonaive and recurrent samples. Notably, we identified CD1d as a biomarker of the OC microenvironment and demonstrated its targeting by invariant natural killer T (iNKT) cells. Overall, our study presents a comprehensive immuno-profiling of OC tumor and microenvironment during disease progression, guiding the development of immunotherapies for OC treatment, especially for recurrent disease.

Spatial Transcriptomics Depict Ligand-Receptor Cross-talk Heterogeneity at the Tumor-Stroma Interface in Long-Term Ovarian Cancer Survivors

Advanced high-grade serous ovarian cancer (HGSC) is an aggressive disease that accounts for 70% of all ovarian cancer deaths. Nevertheless, 15% of patients diagnosed with advanced HGSC survive more than 10 years. The elucidation of predictive markers of these long-term survivors (LTS) could help identify therapeutic targets for the disease, and thus improve patient survival rates. To investigate the stromal heterogeneity of the tumor microenvironment (TME) in ovarian cancer, we used spatial transcriptomics to generate spatially resolved transcript profiles in treatment-naïve advanced HGSC from LTS and short-term survivors (STS) and determined the association between cancer-associated fibroblasts (CAF) heterogeneity and survival in patients with advanced HGSC. Spatial transcriptomics and single-cell RNA-sequencing data were integrated to distinguish tumor and stroma regions, and a computational method was developed to investigate spatially resolved ligand-receptor interactions between various tumor and CAF subtypes in the TME. A specific subtype of CAFs and its spatial location relative to a particular ovarian cancer cell subtype in the TME correlated with long-term survival in patients with advanced HGSC. Also, increased APOE-LRP5 cross-talk occurred at the stroma-tumor interface in tumor tissues from STS compared with LTS. These findings were validated using multiplex IHC. Overall, this spatial transcriptomics analysis revealed spatially resolved CAF-tumor cross-talk signaling networks in the ovarian TME that are associated with long-term survival of patients with HGSC. Further studies to confirm whether such cross-talk plays a role in modulating the malignant phenotype of HGSC and could serve as a predictive biomarker of patient survival are warranted.

SIGNIFICANCE

Generation of spatially resolved gene expression patterns in tumors from patients with ovarian cancer surviving more than 10 years allows the identification of novel predictive biomarkers and therapeutic targets for better patient management. See related commentary by Kelliher and Lengyel, p. 1383.

Computational pathology in ovarian cancer

Histopathologic evaluations of tissue sections are key to diagnosing and managing ovarian cancer. Pathologists empirically assess and integrate visual information, such as cellular density, nuclear atypia, mitotic figures, architectural growth patterns, and higher-order patterns, to determine the tumor type and grade, which guides oncologists in selecting appropriate treatment options. Latent data embedded in pathology slides can be extracted using computational imaging. Computers can analyze digital slide images to simultaneously quantify thousands of features, some of which are visible with a manual microscope, such as nuclear size and shape, while others, such as entropy, eccentricity, and fractal dimensions, are quantitatively beyond the grasp of the human mind. Applications of artificial intelligence and machine learning tools to interpret digital image data provide new opportunities to explore and quantify the spatial organization of tissues, cells, and subcellular structures. In comparison to genomic, epigenomic, transcriptomic, and proteomic patterns, morphologic and spatial patterns are expected to be more informative as quantitative biomarkers of complex and dynamic tumor biology. As computational pathology is not limited to visual data, nuanced subvisual alterations that occur in the seemingly “normal” pre-cancer microenvironment could facilitate research in early cancer detection and prevention. Currently, efforts to maximize the utility of computational pathology are focused on integrating image data with other -omics platforms that lack spatial information, thereby providing a new way to relate the molecular, spatial, and microenvironmental characteristics of cancer. Despite a dire need for improvements in ovarian cancer prevention, early detection, and treatment, the ovarian cancer field has lagged behind other cancers in the application of computational pathology. The intent of this review is to encourage ovarian cancer research teams to apply existing and/or develop additional tools in computational pathology for ovarian cancer and actively contribute to advancing this important field.

Injectable three-dimensional tumor microenvironments to study mechanobiology in ovarian cancer

Epithelial ovarian cancers are among the most aggressive forms of gynecological malignancies. Despite the advent of poly adenosine diphosphate-ribose polymerase (PARP) and checkpoint inhibitors, improvement to patient survival has been modest. Limited in part by clinical translation, beneficial therapeutic strategies remain elusive in ovarian cancers. Although elevated levels of extracellular proteins, including collagens, proteoglycans, and glycoproteins, have been linked to chemoresistance, they are often missing from the processes of drug- development and screening. Biophysical and biochemical signaling from the extracellular matrix (ECM) determine cellular phenotype and affect both tumor progression and therapeutic response. However, many state-of-the-art tumor models fail to mimic the complexities of the tumor microenvironment (TME) and omit key signaling components. In this article, two interpenetrating network (IPN) hydrogel scaffold platforms, comprising of alginate-collagen or agarose-collagen, have been characterized for use as 3D in vitro models of epithelial ovarian cancer ECM. These highly tunable, injection mold compatible, and inexpensive IPNs replicate the critical governing physical and chemical signaling present within the ovarian TME. Additionally, an effective and cell-friendly live-cell retrieval method has been established to recover cells post-encapsulation. Lastly, functional mechanotransduction in ovarian cancers was demonstrated by increasing scaffold stiffness within the 3D in vitro ECM models. With these features, the agarose-collagen and alginate-collagen hydrogels provide a robust TME for the study of mechanobiology in epithelial cancers. STATEMENT OF SIGNIFICANCE: Ovarian cancer is the most lethal gynecologic cancer afflicting women today. Here we present the development, characterization, and validation of 3D interpenetrating platforms to shift the paradigm in standard in vitro modeling. These models help elucidate the roles of biophysical and biochemical cues in ovarian cancer progression. The agarose-collagen and alginate-collagen interpenetrating network (IPN) hydrogels are simple to fabricate, inexpensive, and can be modified to create custom mechanical stiffnesses and concentrations of bio-adhesive motifs. Given that investigations into the roles of biophysical characteristics in ovarian cancers have provided incongruent results, we believe that the IPN platforms will be critically important to uncovering molecular drivers. We also expect these platforms to be broadly applicable to studies involving mechanobiology in solid tumors.

The impact of the spatial heterogeneity of resistant cells and fibroblasts on treatment response

A long-standing practice in the treatment of cancer is that of hitting hard with the maximum tolerated dose to eradicate tumors. This continuous therapy, however, selects for resistant cells, leading to the failure of the treatment. A different type of treatment strategy, adaptive therapy, has recently been shown to have a degree of success in both preclinical xenograft experiments and clinical trials. Adaptive therapy is used to maintain a tumor's volume by exploiting the competition between drug-sensitive and drug-resistant cells with minimum effective drug doses or timed drug holidays. To further understand the role of competition in the outcomes of adaptive therapy, we developed a 2D on-lattice agent-based model. Our simulations show that the superiority of the adaptive strategy over continuous therapy depends on the local competition shaped by the spatial distribution of resistant cells. Intratumor competition can also be affected by fibroblasts, which produce microenvironmental factors that promote cancer cell growth. To this end, we simulated the impact of different fibroblast distributions on treatment outcomes. As a proof of principle, we focused on five types of distribution of fibroblasts characterized by different locations, shapes, and orientations of the fibroblast region with respect to the resistant cells. Our simulation shows that the spatial architecture of fibroblasts modulates tumor progression in both continuous and adaptive therapy. Finally, as a proof of concept, we simulated the outcomes of adaptive therapy of a virtual patient with four metastatic sites composed of different spatial distributions of fibroblasts and drug-resistant cell populations. Our simulation highlights the importance of undetected metastatic lesions on adaptive therapy outcomes.

Visual Analytics for Hypothesis-Driven Exploration in Computational Pathology

Recent advances in computational and algorithmic power are evolving the field of medical imaging rapidly. In cancer research, many new directions are sought to characterize patients with additional imaging features derived from radiology and pathology images. The emerging field of Computational Pathology targets the high-throughput extraction and analysis of the spatial distribution of cells from digital histopathology images. The associated morphological and architectural features allow researchers to quantify and characterize new imaging biomarkers for cancer diagnosis, prognosis, and treatment decisions. However, while the image feature space grows, exploration and analysis become more difficult and ineffective. There is a need for dedicated interfaces for interactive data manipulation and visual analysis of computational pathology and clinical data. For this purpose, we present IIComPath, a visual analytics approach that enables clinical researchers to formulate hypotheses and create computational pathology pipelines involving cohort construction, spatial analysis of image-derived features, and cohort analysis. We demonstrate our approach through use cases that investigate the prognostic value of current diagnostic features and new computational pathology biomarkers.

Digital Pathology-based Study of Cell- and Tissue-level Morphologic Features in Serous Borderline Ovarian Tumor and High-grade Serous Ovarian Cancer

Background:

Serous borderline ovarian tumor (SBOT) and high-grade serous ovarian cancer (HGSOC) are two distinct subtypes of epithelial ovarian tumors, with markedly different biologic background, behavior, prognosis, and treatment. However, the histologic diagnosis of serous ovarian tumors can be subjectively variable and labor-intensive as multiple tumor slides/blocks need to be thoroughly examined to search for these features.

Materials and Methods:

We developed a novel informatics system to facilitate objective and scalable diagnosis screening for SBOT and HGSOC. The system was built upon Groovy scripts and QuPath to enable interactive annotation and data exchange.

Results:

The system was used to successfully detect cellular boundaries and extract an expanded set of cellular features representing cell- and tissue-level characteristics. The performance of cell-level classification for both tumor and stroma cells achieved >90% accuracy. The performance of differentiating HGSOC versus SBOT achieved 91%–95% accuracy for 6485 imaging patches which have sufficient tumor and stroma cells (minimum of ten each) and 97% accuracy for classifying patients when aggregating the results to whole-slide image based on consensus.

Conclusions:

Cellular features digitally extracted from pathological images can be used for cell classification and SBOT v. HGSOC differentiation. Introducing digital pathology into ovarian cancer research could be beneficial to discover potential clinical implications. A larger cohort is required to further evaluate the system.

Expression and prognostic significance of NKD2 in ovarian cancer

PURPOSE

Naked2 (NKD2) is a negative regulator of Wnt signaling pathway and associates with transforming growth factor secretion. The role of NKD2 in ovarian cancer is unknown.

PATIENTS AND METHODS

Gene expression profiles were measured and compared in nine patients by RNA sequencing. NKD2 expressions in ovarian cancer were measured by reverse transcription polymerase chain reaction and western blot. Tissue slides of 79 patients were stained and scored for NKD2 expression. In vitro experiments were conducted to explore the role of NKD2 in ovarian cancer. The prognostic role of NKD2 was evaluated by survival analysis.

RESULTS

NKD2 was upregulated in patients with better survival by mRNA and protein expression. Patients were classified as NKD2-high group (n = 30) and NKD2-low group (n = 49) according to immunohistochemical score. High NKD2 was correlated with lower recurrence rate (P = 0.002) and higher percentage of platinum-sensitive recurrence (P = 0.006). Median progression-free survival was significantly longer for NKD2-high patients than NKD2-low patients (49.1 vs.14.1 months, P < 0.001). Accordingly, there was a significantly difference in terms of overall survival time between two groups (hazard ratio: 3.04; 95% confidence interval: 1.58-5.85, P < 0.001). Multivariate regression suggested that NKD2 was independently prognostic factors in terms of progression-free survival (hazard ratio: 2.91; 95% confidence interval: 1.61-5.27, P < 0.001) and overall survival (hazard ratio: 3.6; 95% confidence interval: 1.80-7.21, P < 0.001). In vitro studies further demonstrated that NKD2 suppressed ovarian cancer cell proliferation, colony formation and cell migration.

CONCLUSION

NKD2 is a novel prognostic marker and could suppress tumor progression in ovarian cancer.

Clinical Implication of Systemic Immune-Inflammation Index and Prognostic Nutritional Index in Skull Base Chordoma Patients

Inflammation associated markers and nutritional indexes are associated with survival, and act as novel prognostic grading systems in patients with cancer, though the role of these markers in chordoma remains unclear. The current study aimed to characterize systemic immune-inflammation index (SII) and prognostic nutritional index (PNI), and their relationship with clinicopathological data and survival in skull base chordoma. Our retrospective study enrolled 183 patients with primary skull base chordoma who received surgical treatment. Clinicopathological data and preoperative blood tests including neutrophil, lymphocyte, platelet counts and albumin level were collected from medical records. Neutrophil lymphocyte ratio (NLR), platelet lymphocyte ratio (PLR), SII, PNI were calculated and the optimal cut-off values of these markers were used for further survival analysis via Kaplan–Meier survival analysis and Cox proportional hazards regression analysis. The value of NLR, PLR, SII, and PNI in skull base chordoma ranged from 0.44–6.48, 45.36–273.94, 113.37–1761.45, and 43.40–70.65, respectively. PNI was significantly correlated with patients' sex (p = 0.005) and age (p = 0.037). SII was positively correlated with NLR and PLR, but negatively correlated with PNI. The median overall survival (OS) time was 74.0 months and Kaplan–Meier survival analysis indicated that all four indexes were associated with OS. Multivariable Cox proportional hazards regression analysis identified that high SII was an independent prognostic factor for poor OS. More importantly, patients with high SII and PNI had the worst outcomes and combined use of SII and PNI increased the predictive ability for patients' survival in skull base chordoma. Our results suggest SII and PNI may be effective prognostic indicators of OS for patients with primary skull base chordoma after surgical resection.

The Relationship Between Tumor-Stroma Ratio, the Immune Microenvironment, and Survival in Patients With Spinal Chordoma

BACKGROUND

Currently, little is known about the clinical relevance of tumor-stroma ratio (TSR) in chordoma and data discussing the relationship between TSR and immune status of chordoma are lacking.

OBJECTIVE

To characterize TSR distribution in spinal chordoma, and investigated its correlation with clinicopathologic or immunological features of patients and outcome.

METHODS

TSR was assessed visually on hematoxylin and eosin-stained sections from 54 tumor specimens by 2 independent pathologists. Multiplex immunofluorescence was used to quantify the expression levels of microvessel density, Ki-67, Brachyury, and tumor as well as stromal PD-L1. Tumor immunity status including the Immunoscore and densities of tumor-infiltrating lymphocytes (TILs) subtypes were obtained from our published data and reanalyzed.

RESULTS

Bland-Altman plot showed no difference between mean TSR derived from the two observers. TSR was positively associated with stromal PD-L1 expression, the Immunoscore and CD3+ as well as CD4+ TILs density, but negatively correlated with tumor microvessel density, Ki-67 index, surrounding muscle invasion by tumor and number of Foxp3+ and PD-1+ TILs. Low TSR independently predicted poor local recurrence-free survival and overall survival. Moreover, patients with low TSR and low Immunoscore chordoma phenotype were associated with the worst survival. More importantly, combined TSR and Immunoscore accurately reflected prognosis and enhanced the ability of TSR or Immunoscore alone for outcome prediction.

CONCLUSION

These data reveal the significant impact of TSR on tumor progression and immunological response of patients. Subsequent use of agents targeting the stroma compartment may be an effective strategy to treat chordoma especially in combination with immune-based drugs.

Why is cancer not more common? A changing microenvironment may help to explain why, and suggests strategies for anti-cancer therapy

One of the great unsolved puzzles in cancer biology is not why cancers occur, but rather explaining why so few cancers occur compared with the theoretical number that could occur, given the number of progenitor cells in the body and the normal mutation rate. We hypothesized that a contributory explanation is that the tumour microenvironment (TME) is not fixed due to factors such as immune cell infiltration, and that this could impair the ability of neoplastic cells to retain a high enough fitness to become a cancer. The TME has implicitly been assumed to be static in most cancer evolution models, and we therefore developed a mathematical model of spatial cancer evolution assuming that the TME, and thus the optimum cancer phenotype, changes over time. Based on simulations, we show how cancer cell populations adapt to diverse changing TME conditions and fitness landscapes. Compared with static TMEs, which generate neutral dynamics, changing TMEs lead to complex adaptations with characteristic spatio-temporal heterogeneity involving variable fitness effects of driver mutations, subclonal mixing, subclonal competition and phylogeny patterns. In many cases, cancer cell populations fail to grow or undergo spontaneous regression, and even extinction. Our analyses predict that cancer evolution in a changing TME is challenging, and can help to explain why cancer is neither inevitable nor as common as expected. Should cancer driver mutations with effects dependent of the TME exist, they are likely to be selected. Anti-cancer prevention and treatment strategies based on changing the TME are feasible and potentially effective.

Organotypic breast tumor model elucidates dynamic remodeling of tumor microenvironment

Fibroblasts are a critical component of tumor microenvironments and associate with cancer cells physically and biochemically during different stages of the disease. Existing cell culture models to study interactions between fibroblasts and cancer cells lack native tumor architecture or scalability. We developed a scalable organotypic model by robotically encapsulating a triple negative breast cancer (TNBC) cell spheroid within a natural extracellular matrix containing dispersed fibroblasts. We utilized an established CXCL12 - CXCR4 chemokine-receptor signaling in breast tumors to validate our model. Using imaging techniques and molecular analyses, we demonstrated that CXCL12-secreting fibroblasts have elevated activity of RhoA/ROCK/myosin light chain-2 pathway and rapidly and significantly contract collagen matrices. Signaling between TNBC cells and CXCL12-producing fibroblasts promoted matrix invasion of cancer cells by activating oncogenic mitogen-activated protein kinase signaling, whereas normal fibroblasts significantly diminished TNBC cell invasiveness. We demonstrated that disrupting CXCL12 - CXCR4 signaling using a molecular inhibitor significantly inhibited invasiveness of cancer cells, suggesting blocking of tumor-stromal interactions as a therapeutic strategy especially for cancers such as TNBC that lack targeted therapies. Our organotypic tumor model mimics native solid tumors, enables modular addition of different stromal cells and extracellular matrix proteins, and allows high throughput compound screening against tumor-stromal interactions to identify novel therapeutics.

A four-factor immune risk score signature predicts the clinical outcome of patients with spinal chordoma

BACKGROUND

Currently, the measurement of immune cells in previous studies is usually subjective, and no immune-based prognostic model has been established for chordoma. In this study, we sought to simultaneously measure tumor-infiltrating lymphocyte (TIL) subtypes in chordoma samples using an objective method and develop an immune risk score (IRS) model for survival prediction.

METHODS

Multiplexed quantitative immunofluorescence staining was used to determine the TIL levels in the tumoral and stromal subareas of 114 spinal chordoma specimens (54 in the training and 60 in the validation cohort) for programmed death-1 (PD-1), CD3, CD8, CD20 (where CD is cluster of differentiation), and FOXP3. Flow cytometry was performed to validate the immunofluorescence assay for lymphocyte measurement on an additional five fresh chordoma specimens. Subsequently, the IRS model was built using the least absolute shrinkage and selection operator (LASSO) Cox regression method.

RESULTS

Flow cytometry and quantitative immunofluorescence showed similar lymphocytic percentages and TIL subpopulation proportions in the fresh tumor specimens. With the training data, the LASSO model identified four immune features for IRS construction: tumoral FOXP3, tumoral PD-1, stromal FOXP3, and stromal CD8. In both cohorts, a high IRS was significantly associated with tumoral programmed cell death-1 ligand 1 expression, Enneking inappropriate tumor resection, and surrounding muscle invasion by tumor. Multivariate Cox regression and stratified analysis in the two cohorts revealed that the IRS was an independent predictor and could effectively separate patients with similar Enneking staging into different risk subgroups, with significantly different survival rates. Further receiver operating characteristic analysis found that the IRS classifier had a better prognostic value than the traditional clinicopathological factors and compensated for the deficiency of Enneking staging for outcome prediction. More importantly, a nomogram based on the IRS and clinical predictors showed adequate performance in estimating disease recurrence and survival of patients.

CONCLUSIONS

These data support the use of the IRS signature as a reliable prognostic tool in spinal chordoma and may facilitate individualized therapy decision making for patients.

Deep learning pathological microscopic features in endemic nasopharyngeal cancer: Prognostic value and protentional role for individual induction chemotherapy

BACKGROUND

To explore the prognostic value and the role for treatment decision of pathological microscopic features in patients with nasopharyngeal carcinoma (NPC) using the method of deep learning.

METHODS

The pathological microscopic features were extracted using the software QuPath (version 0.1.3. Queen's University) in the training cohort (Guangzhou training cohort, n = 843). We used the neural network DeepSurv to analyze the pathological microscopic features (DSPMF) and then classified patients into high-risk and low-risk groups through the time-dependent receiver operating characteristic (ROC). The prognosis accuracy of the pathological feature was validated in a validation cohort (n = 212). The primary endpoint was progression-free survival (PFS).

RESULTS

We found 429 pathological microscopic features in the H&E image. Patients with high-risk scores in the training cohort had shorter 5-year PFS (HR 10.03, 6.06-16.61; P < .0001). The DSPMF (C-index: 0.723) had the higher C-index than the EBV DNA (C-index: 0.612) copies and the N stage (C-index: 0.593). Furthermore, induction chemotherapy (ICT) plus concomitant chemoradiotherapy (CCRT) had better 5-year PFS to those received CCRT (P < .0001) in the high-risk group.

CONCLUSION

The DSPMF is a reliable prognostic tool for survival risk in patients with NPC and might be able to guide the treatment decision.

Analysis of tumour ecological balance reveals resource-dependent adaptive strategies of ovarian cancer

BACKGROUND

Despite treatment advances, there remains a significant risk of recurrence in ovarian cancer, at which stage it is usually incurable. Consequently, there is a clear need for improved patient stratification. However, at present clinical prognosticators remain largely unchanged due to the lack of reproducible methods to identify high-risk patients.

METHODS

In high-grade serous ovarian cancer patients with advanced disease, we spatially define a tumour ecological balance of stromal resource and immune hazard using high-throughput image and spatial analysis of routine histology slides. On this basis an EcoScore is developed to classify tumours by a shift in this balance towards cancer-favouring or inhibiting conditions.

FINDINGS

The EcoScore provides prognostic value stronger than, and independent of, known risk factors. Crucially, the clinical relevance of mutational burden and genomic instability differ under different stromal resource conditions, suggesting that the selective advantage of these cancer hallmarks is dependent on the context of stromal spatial structure. Under a high resource condition defined by a high level of geographical intermixing of cancer and stromal cells, selection appears to be driven by point mutations; whereas, in low resource tumours featured with high hypoxia and low cancer-immune co-localization, selection is fuelled by aneuploidy.

INTERPRETATION

Our study offers empirical evidence that cancer fitness depends on tumour spatial constraints, and presents a biological basis for developing better assessments of tumour adaptive strategies in overcoming ecological constraints including immune surveillance and hypoxia.

Stromal cell ratio based on automated image analysis as a predictor for platinum-resistant recurrent ovarian cancer

Background

Identifying high-risk patients for platinum resistance is critical for improving clinical management of ovarian cancer. We aimed to use automated image analysis of hematoxylin & eosin (H&E) stained sections to identify the association between microenvironmental composition and platinum-resistant recurrent ovarian cancer.

Methods

Ninety-one patients with ovarian cancer containing the data of automated image analysis for H&E histological sections were initially reviewed.

Results

Seventy-one patients with recurrent disease were finally identified. Among 30 patients with high stromal cell ratio, 60% of the patients had platinum-resistant recurrence, which was significantly higher than the rate in patients with low stromal cell ratio (9.80%, P <  0.001). Multivariate logistic regression analysis revealed elevated CA125 level after 3 cycles of chemotherapy (P <  0.001) and high stromal cell ratio (P = 0.002) were the negative predictors of platinum-resistant relapse. The area under the curve (AUC) of receiver operating characteristic (ROC) curves of the models for predicting platinum-resistant recurrence with stromal cell ratio, normalization of CA125 level, and the combination of two parameters were 0.78, 0.79, and 0.89 respectively.

Conclusions

Our results demonstrated stromal cell ratio based on automated image analysis may be a potential predictor for ovarian cancer patients at high risk of platinum-resistant recurrence, and it could improve the predictive value of model when combined with normalization of CA125 level after 3 cycles of chemotherapy.

Microenvironmental niche divergence shapes BRCA1-dysregulated ovarian cancer morphological plasticity

How tumor microenvironmental forces shape plasticity of cancer cell morphology is poorly understood. Here, we conduct automated histology image and spatial statistical analyses in 514 high grade serous ovarian samples to define cancer morphological diversification within the spatial context of the microenvironment. Tumor spatial zones, where cancer cell nuclei diversify in shape, are mapped in each tumor. Integration of this spatially explicit analysis with omics and clinical data reveals a relationship between morphological diversification and the dysregulation of DNA repair, loss of nuclear integrity, and increased disease mortality. Within the Immunoreactive subtype, spatial analysis further reveals significantly lower lymphocytic infiltration within diversified zones compared with other tumor zones, suggesting that even immune-hot tumors contain cells capable of immune escape. Our findings support a model whereby a subpopulation of morphologically plastic cancer cells with dysregulated DNA repair promotes ovarian cancer progression through positive selection by immune evasion.

Microenvironmental niche divergence shapes BRCA1-dysregulated ovarian cancer morphological plasticity

How tumor microenvironmental forces shape plasticity of cancer cell morphology is poorly understood. Here, we conduct automated histology image and spatial statistical analyses in 514 high grade serous ovarian samples to define cancer morphological diversification within the spatial context of the microenvironment. Tumor spatial zones, where cancer cell nuclei diversify in shape, are mapped in each tumor. Integration of this spatially explicit analysis with omics and clinical data reveals a relationship between morphological diversification and the dysregulation of DNA repair, loss of nuclear integrity, and increased disease mortality. Within the Immunoreactive subtype, spatial analysis further reveals significantly lower lymphocytic infiltration within diversified zones compared with other tumor zones, suggesting that even immune-hot tumors contain cells capable of immune escape. Our findings support a model whereby a subpopulation of morphologically plastic cancer cells with dysregulated DNA repair promotes ovarian cancer progression through positive selection by immune evasion.

Microenvironmental niche divergence shapes BRCA1-dysregulated ovarian cancer morphological plasticity

How tumor microenvironmental forces shape plasticity of cancer cell morphology is poorly understood. Here, we conduct automated histology image and spatial statistical analyses in 514 high grade serous ovarian samples to define cancer morphological diversification within the spatial context of the microenvironment. Tumor spatial zones, where cancer cell nuclei diversify in shape, are mapped in each tumor. Integration of this spatially explicit analysis with omics and clinical data reveals a relationship between morphological diversification and the dysregulation of DNA repair, loss of nuclear integrity, and increased disease mortality. Within the Immunoreactive subtype, spatial analysis further reveals significantly lower lymphocytic infiltration within diversified zones compared with other tumor zones, suggesting that even immune-hot tumors contain cells capable of immune escape. Our findings support a model whereby a subpopulation of morphologically plastic cancer cells with dysregulated DNA repair promotes ovarian cancer progression through positive selection by immune evasion.

Platinum sensitivity of ovarian cancer cells does not influence their ability to induce M2-type macrophage polarization

PROBLEM

Development of platinum resistance in ovarian cancer is mediated by both cancer cells and tumor microenvironment. Activation of epithelial-mesenchymal transition program in cancer cells may lead to enrichment for resistant clones. These processes can be affected by tumor-associated macrophages, a highly plastic population of cells that participate in tumor progression and response to treatment by shaping the microenvironment. We aimed to study how platinum resistance influences the crosstalk between macrophages and ovarian cancer cells.

METHOD OF STUDY

Using cisplatin-sensitive ovarian cancer cell line A2780, we developed and characterized cisplatin-resistant A2780Cis and cisplatin and doxorubicin co-resistant A2780Dox cell lines. Next, we set up an indirect coculture system with THP-1 cell line-derived M0-type-, M1-type- and M2-type-like polarized macrophages. We monitored the expression of genes associated with cellular stemness, multidrug resistance, and epithelial-mesenchymal transition in cancer cells, and expression profile of M1/M2 markers in macrophages.

RESULTS

Development of drug resistance in ovarian cancer cell lines was accompanied by increased migration, clonogenicity, and upregulated expression of transcription factors, associated with cellular stemness and epithelial-mesenchymal transition. Upon coculture, we noted that the most relevant changes in gene expression profile occurred in A2780 cells. Moreover, M0- and M1-type macrophages, but not M2-type macrophages, showed significant transcriptional alterations.

CONCLUSION

Our results provide the evidence for bidirectional interplay between cancer cells and macrophages. Independent of platinum resistance status, ovarian cancer cells polarize macrophages toward M2-like type, whereas macrophages induce epithelial-mesenchymal transition and stemness-related gene expression profile in cisplatin-sensitive, but not cisplatin-resistant cancer cells.

Image-based surrogate biomarkers for molecular subtypes of colorectal cancer

Motivation

Whole genome expression profiling of large cohorts of different types of cancer led to the identification of distinct molecular subcategories (subtypes) that may partially explain the observed inter-tumoral heterogeneity. This is also the case of colorectal cancer (CRC) where several such categorizations have been proposed. Despite recent developments, the problem of subtype definition and recognition remains open, one of the causes being the intrinsic heterogeneity of each tumor, which is difficult to estimate from gene expression profiles. However, one of the observations of these studies indicates that there may be links between the dominant tumor morphology characteristics and the molecular subtypes. Benefiting from a large collection of CRC samples, comprising both gene expression and histopathology images, we investigated the possibility of building image-based classifiers able to predict the molecular subtypes. We employed deep convolutional neural networks for extracting local descriptors which were then used for constructing a dictionary-based representation of each tumor sample. A set of support vector machine classifiers were trained to solve different binary decision problems, their combined outputs being used to predict one of the five molecular subtypes.

Results

A hierarchical decomposition of the multi-class problem was obtained with an overall accuracy of 0.84 (95%CI=0.79-0.88). The predictions from the image-based classifier showed significant prognostic value similar to their molecular counterparts.

Contact

popovici@iba.muni.cz.

Availability and Implementation

Source code used for the image analysis is freely available from https://github.com/higex/qpath .

Supplementary information

Supplementary data are available at Bioinformatics online.

Similarity and diversity of the tumor microenvironment in multiple metastases: critical implications for overall and progression-free survival of high-grade serous ovarian cancer

The tumor microenvironment is pivotal in influencing cancer progression and metastasis. Different cells co-exist with high spatial diversity within a patient, yet their combinatorial effects are poorly understood. We investigate the similarity of the tumor microenvironment of 192 local metastatic lesions in 61 ovarian cancer patients. An ecologically inspired measure of microenvironmental diversity derived from multiple metastasis sites is correlated with clinicopathological characteristics and prognostic outcome. We demonstrate a high accuracy of our automated analysis across multiple sites. A low level of similarity in microenvironmental composition is observed between ovary tumor and corresponding local metastases (stromal ratio r = 0.30, lymphocyte ratio r = 0.37). We identify a new measure of microenvironmental diversity derived from Shannon entropy that is highly predictive of poor overall (p = 0.002, HR = 3.18, 95% CI = 1.51-6.68) and progression-free survival (p = 0.0036, HR = 2.83, 95% CI = 1.41-5.7), independent of and stronger than clinical variables, subtype stratifications based on single cell types alone and number of sites. Although stromal influence in ovary tumors is known to have significant clinical implications, our findings reveal an even stronger impact orchestrated by diverse cell types. Quantitative histology-based measures can further enable objective selection of patients who are in urgent need of new therapeutic strategies such as combinatorial treatments targeting heterogeneous tumor microenvironment.

Size-Based Differentiation of Cancer and Normal Cells by a Particle Size Analyzer Assisted by a Cell-Recognition PC Software

Detection of anomalous cells such as cancer cells from normal blood cells has the potential to contribute greatly to cancer diagnosis and therapy. Conventional methods for the detection of cancer cells are usually tedious and cumbersome. Herein, we report on the use of a particle size analyzer for the convenient size-based differentiation of cancer cells from normal cells. Measurements made using a particle size analyzer revealed that size parameters for cancer cells are significantly greater (e.g., inner diameter and width) than the corresponding values for normal cells (white blood cells (WBC), lymphocytes and splenocytes), with no significant difference in shape parameters (e.g., circularity and convexity). The inner diameter of many cancer cell lines is greater than 10 µm, in contrast to normal cells. For the detection of WBC having similar size to that of cancer cells, we developed a PC software "Cancer Cell Finder" that differentiates them from cancer cells based on brightness stationary points on a cell surface. Furthermore, the aforementioned method was validated for cancer cell/clusters detection in spiked mouse blood samples (a B16 melanoma mouse xenograft model) and circulating tumor cell cluster-like particles in the cat and dog (diagnosed with cancer) blood samples. These results provide insights into the possible applicability of the use of a particle size analyzer in conjunction with PC software for the convenient detection of cancer cells in experimental and clinical samples for theranostics.

A novel machine learning approach reveals latent vascular phenotypes predictive of renal cancer outcome

Gene expression signatures are commonly used as predictive biomarkers, but do not capture structural features within the tissue architecture. Here we apply a 2-step machine learning framework for quantitative imaging of tumor vasculature to derive a spatially informed, prognostic gene signature. The trained algorithms classify endothelial cells and generate a vascular area mask (VAM) in H&E micrographs of clear cell renal cell carcinoma (ccRCC) cases from The Cancer Genome Atlas (TCGA). Quantification of VAMs led to the discovery of 9 vascular features (9VF) that predicted disease-free-survival in a discovery cohort (n = 64, HR = 2.3). Correlation analysis and information gain identified a 14 gene expression signature related to the 9VF's. Two generalized linear models with elastic net regularization (14VF and 14GT), based on the 14 genes, separated independent cohorts of up to 301 cases into good and poor disease-free survival groups (14VF HR = 2.4, 14GT HR = 3.33). For the first time, we successfully applied digital image analysis and targeted machine learning to develop prognostic, morphology-based, gene expression signatures from the vascular architecture. This novel morphogenomic approach has the potential to improve previous methods for biomarker development.

Automated analysis of co-localized protein expression in histologic sections of prostate cancer

An automated approach based on routinely-processed, whole-slide immunohistochemistry (IHC) was implemented to study co-localized protein expression in tissue samples. Expression of two markers was chosen to represent stromal (CD31) and epithelial (Ki-67) compartments in prostate cancer. IHC was performed on whole-slide sections representing low-, intermediate-, and high-grade disease from 15 patients. The automated workflow was developed using a training set of regions-of-interest in sequential tissue sections. Protein expression was studied on digital representations of IHC images across entire slides representing formalin-fixed paraffin embedded blocks. Using the training-set, the known association between Ki-67 and Gleason grade was confirmed. CD31 expression was more heterogeneous across samples and remained invariant with grade in this cohort. Interestingly, the Ki-67/CD31 ratio was significantly increased in high (Gleason ≥ 8) versus low/intermediate (Gleason ≤7) samples when assessed in the training-set and the whole-tissue block images. Further, the feasibility of the automated approach to process Tissue Microarray (TMA) samples in high throughput was evaluated. This work establishes an initial framework for automated analysis of co-localized protein expression and distribution in high-resolution digital microscopy images based on standard IHC techniques. Applied to a larger sample population, the approach may help to elucidate the biologic basis for the Gleason grade, which is the strongest, single factor distinguishing clinically aggressive from indolent prostate cancer.

Relevance of deep learning to facilitate the diagnosis of HER2 status in breast cancer

Tissue biomarker scoring by pathologists is central to defining the appropriate therapy for patients with cancer. Yet, inter-pathologist variability in the interpretation of ambiguous cases can affect diagnostic accuracy. Modern artificial intelligence methods such as deep learning have the potential to supplement pathologist expertise to ensure constant diagnostic accuracy. We developed a computational approach based on deep learning that automatically scores HER2, a biomarker that defines patient eligibility for anti-HER2 targeted therapies in breast cancer. In a cohort of 71 breast tumour resection samples, automated scoring showed a concordance of 83% with a pathologist. The twelve discordant cases were then independently reviewed, leading to a modification of diagnosis from initial pathologist assessment for eight cases. Diagnostic discordance was found to be largely caused by perceptual differences in assessing HER2 expression due to high HER2 staining heterogeneity. This study provides evidence that deep learning aided diagnosis can facilitate clinical decision making in breast cancer by identifying cases at high risk of misdiagnosis.

Environmental factors in breast cancer invasion: a mathematical modelling review

This review presents a brief overview of breast cancer, focussing on its heterogeneity and the role of mathematical modelling and simulation in teasing apart the underlying biophysical processes. Following a brief overview of the main known pathophysiological features of ductal carcinoma, attention is paid to differential equation-based models (both deterministic and stochastic), agent-based modelling, multi-scale modelling, lattice-based models and image-driven modelling. A number of vignettes are presented where these modelling approaches have elucidated novel aspects of breast cancer dynamics, and we conclude by offering some perspectives on the role mathematical modelling can play in understanding breast cancer development, invasion and treatment therapies.

When machine vision meets histology: A comparative evaluation of model architecture for classification of histology sections

Classification of histology sections in large cohorts, in terms of distinct regions of microanatomy (e.g., stromal) and histopathology (e.g., tumor, necrosis), enables the quantification of tumor composition, and the construction of predictive models of genomics and clinical outcome. To tackle the large technical variations and biological heterogeneities, which are intrinsic in large cohorts, emerging systems utilize either prior knowledge from pathologists or unsupervised feature learning for invariant representation of the underlying properties in the data. However, to a large degree, the architecture for tissue histology classification remains unexplored and requires urgent systematical investigation. This paper is the first attempt to provide insights into three fundamental questions in tissue histology classification: I. Is unsupervised feature learning preferable to human engineered features? II. Does cellular saliency help? III. Does the sparse feature encoder contribute to recognition? We show that (a) in I, both Cellular Morphometric Feature and features from unsupervised feature learning lead to superior performance when compared to SIFT and [Color, Texture]; (b) in II, cellular saliency incorporation impairs the performance for systems built upon pixel-/patch-level features; and (c) in III, the effect of the sparse feature encoder is correlated with the robustness of features, and the performance can be consistently improved by the multi-stage extension of systems built upon both Cellular Morphmetric Feature and features from unsupervised feature learning. These insights are validated with two cohorts of Glioblastoma Multiforme (GBM) and Kidney Clear Cell Carcinoma (KIRC).

Multi-class texture analysis in colorectal cancer histology

Automatic recognition of different tissue types in histological images is an essential part in the digital pathology toolbox. Texture analysis is commonly used to address this problem; mainly in the context of estimating the tumour/stroma ratio on histological samples. However, although histological images typically contain more than two tissue types, only few studies have addressed the multi-class problem. For colorectal cancer, one of the most prevalent tumour types, there are in fact no published results on multiclass texture separation. In this paper we present a new dataset of 5,000 histological images of human colorectal cancer including eight different types of tissue. We used this set to assess the classification performance of a wide range of texture descriptors and classifiers. As a result, we found an optimal classification strategy that markedly outperformed traditional methods, improving the state of the art for tumour-stroma separation from 96.9% to 98.6% accuracy and setting a new standard for multiclass tissue separation (87.4% accuracy for eight classes). We make our dataset of histological images publicly available under a Creative Commons license and encourage other researchers to use it as a benchmark for their studies.