The proportion of tumor-stroma as a strong prognosticator for stage II and III colon cancer patients: validation in the VICTOR trial

Artificial intelligence quantified tumour-stroma ratio is an independent predictor for overall survival in resectable colorectal cancer

Background

An artificial intelligence method could accelerate the clinical implementation of tumour-stroma ratio (TSR), which has prognostic relevance in colorectal cancer (CRC). We, therefore, developed a deep learning model for the fully automated TSR quantification on routine haematoxylin and eosin (HE) stained whole-slide images (WSI) and further investigated its prognostic validity for patient stratification.

Methods

We trained a convolutional neural network (CNN) model using transfer learning, with its nine-class tissue classification performance evaluated in two independent test sets. Patch-level segmentation on WSI HE slides was performed using the model, with TSR subsequently derived. A discovery (N=499) and validation cohort (N=315) were used to evaluate the prognostic value of TSR for overall survival (OS).

Findings

The CNN-quantified TSR was a prognostic factor, independently of other clinicopathologic characteristics, with stroma-high associated with reduced OS in the discovery (HR 1.72, 95% CI 1.24-2.37, P=0.001) and validation cohort (2.08, 1.26-3.42, 0.004). Integrating TSR into a Cox model with other risk factors showed improved prognostic capability.

Interpretation

We developed a deep learning model to quantify TSR based on histologic WSI of CRC and demonstrated its prognostic validity for patient stratification for OS in two independent CRC patient cohorts. This fully automatic approach allows for the objective and standardised application while reducing pathologists' workload. Thus, it can potentially be of significant aid in clinical prognosis prediction and decision-making.

Funding

National Key Research and Development Program of China, National Science Fund for Distinguished Young Scholar, and National Science Foundation for Young Scientists of China.

Histological phenotypic subtypes predict recurrence risk and response to adjuvant chemotherapy in patients with stage III colorectal cancer

Histological 'phenotypic subtypes' that classify patients into four groups (immune, canonical, latent and stromal) have previously been demonstrated to stratify survival in a stage I-III colorectal cancer (CRC) pilot cohort. However, clinical utility has not yet been validated. Therefore, this study assessed prognostic value of these subtypes in additional patient cohorts along with associations with risk of recurrence and response to chemotherapy. Two independent stage I-III CRC patient cohorts (internal and external cohort) were utilised to investigate phenotypic subtypes. The primary endpoint was disease-free survival (DFS) and the secondary endpoint was recurrence risk (RR). Stage II-III patients, from the SCOT adjuvant chemotherapy trial, were utilised to further validate prognostic value and for exploratory analysis assessing associations with adjuvant chemotherapy. In an 893-patient internal cohort, phenotypic subtype independently associated with DFS (p = 0.025) and this was attenuated in stage III patients (p = 0.020). Phenotypic subtype also independently associated with RR (p < 0.001) in these patients. In a 146-patient external cohort, phenotypic subtype independently stratified patients by DFS (p = 0.028), validating their prognostic value. In 1343 SCOT trial patients, the effect of treatment type significantly depended on phenotypic subtype (pinteraction = 0.011). Phenotypic subtype independently associated with DFS in stage III patients receiving FOLFOX (p = 0.028). Furthermore, the immune subtype significantly associated with better response to FOLFOX compared to CAPOX adjuvant chemotherapy in stage III patients (p = 0.013). In conclusion, histological phenotypic subtypes are an effective prognostic classification in patients with stage III CRC that associates with risk of recurrence and response to FOLFOX adjuvant chemotherapy.

Multiplex Cellular Communities in Multi-Gigapixel Colorectal Cancer Histology Images for Tissue Phenotyping

In computational pathology, automated tissue phenotyping in cancer histology images is a fundamental tool for profiling tumor microenvironments. Current tissue phenotyping methods use features derived from image patches which may not carry biological significance. In this work, we propose a novel multiplex cellular community-based algorithm for tissue phenotyping integrating cell-level features within a graph-based hierarchical framework. We demonstrate that such integration offers better performance compared to prior deep learning and texture-based methods as well as to cellular community based methods using uniplex networks. To this end, we construct celllevel graphs using texture, alpha diversity and multi-resolution deep features. Using these graphs, we compute cellular connectivity features which are then employed for the construction of a patch-level multiplex network. Over this network, we compute multiplex cellular communities using a novel objective function. The proposed objective function computes a low-dimensional subspace from each cellular network and subsequently seeks a common low-dimensional subspace using the Grassmann manifold. We evaluate our proposed algorithm on three publicly available datasets for tissue phenotyping, demonstrating a significant improvement over existing state-of-the-art methods.

Spatial expression analyses of the putative oncogene ciRS-7 in cancer reshape the microRNA sponge theory

Circular RNAs (circRNAs) have recently gained substantial attention in the cancer research field where most, including the putative oncogene ciRS-7 (CDR1as), have been proposed to function as competitive endogenous RNAs (ceRNAs) by sponging specific microRNAs. Here, we report the first spatially resolved cellular expression patterns of ciRS-7 in colon cancer and show that ciRS-7 is completely absent in the cancer cells, but highly expressed in stromal cells within the tumor microenvironment. Additionally, our data suggest that this generally apply to classical oncogene-driven adenocarcinomas, but not to other cancers, including malignant melanoma. Moreover, we find that correlations between circRNA and mRNA expression, which are commonly interpreted as evidence of a ceRNA function, can be explained by different cancer-to-stromal cell ratios among the studied tumor specimens. Together, these results have wide implications for future circRNA studies and highlight the importance of spatially resolving expression patterns of circRNAs proposed to function as ceRNAs.

The Proportion of Tumour-Stroma in Metastatic Lymph Nodes is An Accurately Prognostic Indicator of Poor Survival for Advanced-Stage Colon Cancers

The importance of tumour microenvironment in tumour behaviour has now become clearer. This study aimed to determine the prognostic effect of the proportion of tumour-stroma (PTS) in metastatic lymph nodes of advanced-stage colon cancers (CCs). We investigated PTS in positive lymph nodes of stage III-IV CC patients who underwent surgical treatment between 2004 and 2014. We used a standard approach in methodology. PTS was significantly associated with prognostic factors in the metastatic lymph nodes (perineural invasion [p = 0.031], lymphatic invasion [p = 0.032], invasive margin [p = 0.043], advanced pT [p = 0.020], and margin involvement [p = 0.034]). In addition, the correlations between PTS estimates (R = 0.704 to 0.617, p < 0.001), the reproducibility of the research (Κappa = 0.72-0.68) and the usefulness of the cut-off value (ROC: 50.33%; AUC = 0.752 [0.667-0.857]) were successful. In univariate analysis, 5-year survival was poor for RFS (p < 0.001), OS (p = 0.001) and LR (p = 0.013) in high PTS patients. Multivariate analysis confirmed that high PTS was an independent worse parameter for RFS (HR = 1.32, 95% CI: 1.17-2.55, p = 0.001) and OS (HR = 1.37, 95% CI: 1.25-1 - 2.56, p = 0.009). In this study, we showed that high PTS in metastatic lymph nodes was a successful prognostic marker for advanced-stage CCs. Also, the standard approach we used for the methodology was successful.

Prognostic Significance of the Tumor-Stromal Ratio in Invasive Breast Cancer and a Proposal of a New Ts-TNM Staging System

Background

Previous studies have demonstrated that the tumor-stromal ratio (TSR) was an independent prognostic factor in several types of carcinomas. This study aimed at exploring the prognostic significance of the TSR in invasive breast cancer using immunohistochemistry (IHC)-stained tissue microarrays (TMAs) and integrating the TSR into the traditional tumor-node-metastasis (TNM) staging system.

Methods

The prepared 7 TMAs containing 240 patients with 480 invasive BC specimens were stained with cytokeratin (CK) by the IHC staining method. The ratio of tumor cells and stromal cells was visually assessed. TSR > 1 and TSR ≤ 1 were categorized as the high TSR (low stroma) and low TSR (high stroma) groups, respectively, and the prognostic value of the TSR at 5-year disease-free survival (5-DFS) was analyzed. A new Ts-TNM (tumor stroma-tumor-node-metastasis) staging system was established and assessed.

Results

IHC staining of CK could specifically label tumor cells with clear contrast, making it easy to manually assess TSR. High TSR (low stroma) and low TSR (high stroma) were observed in 52.5% (n = 126) and 47.5 (n = 114) of the cases, according to the division of value 1. A Kaplan-Meier analysis showed that patients in the low TSR group had a worse 5-DFS compared with patients in the high TSR group (P=0.022). Multivariable analysis indicated that the T stage (P=0.014), N status (P < 0.001), histological grade (P < 0.001), estrogen receptor status (P=0.015), and TSR (P=0.011) were independent prognostic factors of invasive BC patients. The new Ts-TNM staging system combining TSR, tumor staging, lymph node status, and metastasis staging was established. The receiver operating characteristic (ROC) curve analysis demonstrated that the ability of the Ts-TNM staging system to predict recurrence was not lower than that of the TNM staging system.

Conclusions

This study confirms that the TSR is a prognostic indicator for invasive breast cancer. The Ts-TNM staging system containing stromal and tumor information may optimize risk stratification for invasive breast cancer.

Cellular community detection for tissue phenotyping in colorectal cancer histology images

Classification of various types of tissue in cancer histology images based on the cellular compositions is an important step towards the development of computational pathology tools for systematic digital profiling of the spatial tumor microenvironment. Most existing methods for tissue phenotyping are limited to the classification of tumor and stroma and require large amount of annotated histology images which are often not available. In the current work, we pose the problem of identifying distinct tissue phenotypes as finding communities in cellular graphs or networks. First, we train a deep neural network for cell detection and classification into five distinct cellular components. Considering the detected nuclei as nodes, potential cell-cell connections are assigned using Delaunay triangulation resulting in a cell-level graph. Based on this cell graph, a feature vector capturing potential cell-cell connection of different types of cells is computed. These feature vectors are used to construct a patch-level graph based on chi-square distance. We map patch-level nodes to the geometric space by representing each node as a vector of geodesic distances from other nodes in the network and iteratively drifting the patch nodes in the direction of positive density gradients towards maximum density regions. The proposed algorithm is evaluated on a publicly available dataset and another new large-scale dataset consisting of 280K patches of seven tissue phenotypes. The estimated communities have significant biological meanings as verified by the expert pathologists. A comparison with current state-of-the-art methods reveals significant performance improvement in tissue phenotyping.

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.

Artificial intelligence-guided tissue analysis combined with immune infiltrate assessment predicts stage III colon cancer outcomes in PETACC08 study

OBJECTIVE

Diagnostic tests, such as Immunoscore, predict prognosis in patients with colon cancer. However, additional prognostic markers could be detected on pathological slides using artificial intelligence tools.

DESIGN

We have developed a software to detect colon tumour, healthy mucosa, stroma and immune cells on CD3 and CD8 stained slides. The lymphocyte density and surface area were quantified automatically in the tumour core (TC) and invasive margin (IM). Using a LASSO algorithm, DGMate (DiGital tuMor pArameTErs), we detected digital parameters within the tumour cells related to patient outcomes.

RESULTS

Within the dataset of 1018 patients, we observed that a poorer relapse-free survival (RFS) was associated with high IM stromal area (HR 5.65; 95% CI 2.34 to 13.67; p<0.0001) and high DGMate (HR 2.72; 95% CI 1.92 to 3.85; p<0.001). Higher CD3+ TC, CD3+ IM and CD8+ TC densities were significantly associated with a longer RFS. Analysis of variance showed that CD3+ TC yielded a similar prognostic value to the classical CD3/CD8 Immunoscore (p=0.44). A combination of the IM stromal area, DGMate and CD3, designated 'DGMuneS', outperformed Immunoscore when used in estimating patients' prognosis (C-index=0.601 vs 0.578, p=0.04) and was independently associated with patient outcomes following Cox multivariate analysis. A predictive nomogram based on DGMuneS and clinical variables identified a group of patients with less than 10% relapse risk and another group with a 50% relapse risk.

CONCLUSION

These findings suggest that artificial intelligence can potentially improve patient care by assisting pathologists in better defining stage III colon cancer patients' prognosis.

Tumor proportion in colon cancer: results from a semiautomatic image analysis approach

The tumor stroma ratio (TSR) is a promising prognostic biomarker in colon cancer, which could provide additional risk stratification for therapy adaption. The objective of our study was the investigation of the prognostic significance of TSR at different tumor sites in a simple semiautomatic approach with the open-source program ImageJ. We investigated 206 pT3 and pT4 adenocarcinomas of no special type. According to our established thresholds, 31 tumors (15%) were classified as low tumor proportion (TP) (≤ 15% TP), 42 tumors (20%) were classified as high TP (≥ 54% TP), and 133 tumors (65%) were classified as medium TP. High and low TP were associated with an adverse overall survival in comparison to medium TP (p = 0.001 and p = 0.03). Furthermore, the TP was an independent risk factor of occurrence of distant metastasis next to T status, microsatellite status, and tumor budding. The 5-year survival rate was 49% in patients with high TP, 48% in patients with low TP, and 68% in patients with medium TP (p = 0.042, n = 160). Patients with a high TP had less often tumor budding (p = 0.012), lymphovascular invasion (p = 0.049), and less harvested lymph nodes (p = 0.042) in comparison to low TP tumors. The results provide first evidence that a high tumor proportion/low stroma proportion is also associated with an adverse prognosis and that this subgroup might be difficult to identify with other classical histopathologic characteristics that are linked to an adverse prognosis.

Bidirectional tumor/stroma crosstalk promotes metastasis in mesenchymal colorectal cancer

Patients with the mesenchymal subtype colorectal cancer (CRC) have a poor prognosis, in particular patients with stroma-rich tumors and aberrant SMAD4 expression. We hypothesized that interactions between SMAD4-deficient CRC cells and cancer-associated fibroblasts provide a biological explanation. In transwell invasion assays, fibroblasts increased the invasive capacity of SMAD4-deficient HT29 CRC cells, but not isogenic SMAD4-proficient HT29 cells. A TGF-β/BMP-specific array showed BMP2 upregulation by fibroblasts upon stimulation with conditioned medium from SMAD4-deficient CRC cells, while also stimulating their invasion. In a mouse model for experimental liver metastasis, the co-injection of fibroblasts increased metastasis formation of SMAD4-deficient CRC cells (p = 0.02) but not that of SMAD4-proficient CRC cells. Significantly less metastases were seen in mice co-injected with BMP2 knocked-down fibroblasts. Fibroblast BMP2 expression seemed to be regulated by TRAIL, a factor overexpressed in SMAD4-deficient CRC cells. In a cohort of 146 stage III CRC patients, we showed that patients with a combination of high stromal BMP2 expression and the loss of tumor SMAD4 expression had a significantly poorer overall survival (HR 2.88, p = 0.04). Our results suggest the existence of a reciprocal loop in which TRAIL from SMAD4-deficient CRC cells induces BMP2 in fibroblasts, which enhances CRC invasiveness and metastasis.

The prognostic value of the tumor-stroma ratio is most discriminative in patients with grade III or triple-negative breast cancer

The tumor-stroma ratio (TSR) was evaluated as a promising parameter for breast cancer prognostication in clinically relevant subgroups of patients. The TSR was assessed on hematoxylin and eosin-stained tissue slides of 1,794 breast cancer patients from the Nottingham City Hospital. An independent second cohort of 737 patients from the Netherlands Cancer Institute to Antoni van Leeuwenhoek was used for evaluation. In the Nottingham Breast Cancer series, the TSR was an independent prognostic parameter for recurrence-free survival (RFS; HR 1.35, 95% CI 1.10-1.66, p = 0.004). The interaction term was statistically significant for grade and triple-negative status. Multivariate Cox regression analysis showed a more pronounced effect of the TSR for RFS in grade III tumors (HR 1.89, 95% CI 1.43-2.51, p < 0.001) and triple-negative tumors (HR 1.86, 95% CI 1.10-3.14, p = 0.020). Comparable hazard ratios and confidence intervals were observed for grade and triple-negative status in the ONCOPOOL study. The prognostic value of TSR was not modified by age, tumor size, histology, estrogen receptor status, progesterone receptor status, human epidermal growth factor receptor 2 status or lymph node status. In conclusion, patients with a stroma-high tumor had a worse prognosis compared to patients with a stroma-low tumor. The prognostic value of the TSR is most discriminative in grade III tumors and triple-negative tumors. The TSR was not modified by other clinically relevant parameters making it a potential factor to be included for improved risk stratification.

Morphological consistency of desmoplastic reactions between the primary colorectal cancer lesion and associated metastatic lesions

Some preceding studies show that a desmoplastic reaction (DR) at the invasive front of a primary tumor (DR^primary) is a promising prognostic indicator, and its histologic classification is an independent prognostic factor in patients with colorectal cancer (CRC). Although DR is observed in metastatic lesions, such as those in lymph nodes (DR^LN) and the liver (DR^liver), the association between DR^primary and DR of metastatic lesions is unclear. We investigated whether DR^LN and DR^liver could be categorized in the same manner as DR^primary and examined the features and prognostic implications of DR^LN and DR^liver. In this study, we evaluated 363 patients with metastases to lymph nodes and 45 patients with synchronous liver metastases who underwent curative resection. DR^LN and DR^liver statuses were classified as mature, intermediate, or immature, based on keloid-like collagen and myxoid stroma appearances in the metastatic lesions. Overall, 109, 106, and 148 patients had mature, intermediate, and immature DR^LN, respectively; in total, 5, 21, and 19 patients had mature, intermediate, and immature DR^liver, respectively. DR^LN and DR^primary (spearman's rho = 0.43, P < 0.0001) and DR^liver and DR^primary (spearman's rho = 0.40, P = 0.0069) were each significantly correlated. The 5-year relapse-free survival (RFS) after surgery was 67.7% for mature/intermediate DR^LN and 52.9% for immature DR^LN; the 5-year RFS after hepatectomy was 11.5% for mature/intermediate DR^liver and 5.6% for immature DR^liver. In conclusion, DR^LN and DR^liver may be classified in the same manner as DR^primary; morphological consistency of DR was observed between primary and metastatic lesions.

Prognostic value of DNA ploidy and automated assessment of stroma fraction in prostate cancer

The combination of DNA ploidy and automatically estimated stroma fraction has been shown to correlate with recurrence and cancer death in colorectal cancer. We aimed to extend this observation and evaluate the prognostic importance of this combined marker in prostate cancer. DNA ploidy status was determined by image cytometry and the stroma fraction was estimated automatically on hematoxylin and eosin stained sections in three tumor samples from each patient to account for tumor heterogeneity. The optimal threshold for low (≤56%) and high (>56%) stroma fraction was identified in a discovery cohort (n = 253). The combined marker was validated in an independent patient cohort (n = 259) with biochemical recurrence as endpoint. The combined marker predicted biochemical recurrence independently in the validation cohort. Multivariable analysis showed that the highest risk of recurrence was observed for patients with samples that had both non-diploid ploidy status and a high stroma fraction (hazard ratio: 2.51, 95% confidence interval: 1.18-5.34). In conclusion, we suggest the combination of DNA ploidy and automatically estimated stroma fraction as a prognostic marker for the risk stratification of prostate cancer patients. It may also be a potential generic marker as concurrent results have been described in colorectal cancer.

The tumor-stroma ratio is an independent predictor of survival in patients with 2018 FIGO stage IIIC squamous cell carcinoma of the cervix following primary radical surgery

OBJECTIVE

To determine the value of the tumor-stroma ratio (TSR) while identifying prognostic factors in patients with 2018 International Federation of Gynecology and Obstetrics (FIGO) stage IIIC squamous cell carcinoma of the cervix following primary radical surgery.

METHODS

Three hundred eighty-four patients with node-positive squamous cell carcinoma of the cervix (2018 FIGO stage IIIC) who underwent radical surgery between January 2005 and December 2016 were included in this retrospective study. The TSRs were assessed on hematoxylin and eosin-stained tumor slides and classified as stroma-low (<50% stroma) or stroma-high (≥50% stroma).

RESULTS

Sixty-seven patients were categorized as stroma-high; they had shorter disease-free survival (DFS) and overall survival (OS) periods than did their stroma-low counterparts. On multivariate analysis, a tumor size ≥4 cm, ≥3 metastatic lymph nodes, and stroma-high status were independent predictors of shorter DFS and OS. These factors were incorporated into a prognostic scoring system in which patients were categorized into low- (score 0), intermediate- (score 1), and high-risk (scores 2-3) groups. The scoring system differentiated DFS and OS well (C-index = 0.65, 95% confidence interval, 0.59-0.72; and C-index = 0.65, 95% confidence interval, 0.59-0.72, respectively).

CONCLUSIONS

The TSR is an independent prognostic factor, and our prognostic scoring system that incorporates this parameter exhibits good discriminative ability for both recurrence and survival in patients with 2018 FIGO stage IIIC cervical cancer after radical surgery. The TSR is a potentially novel clinicopathological variable for predicting the prognoses of these patients contingent on the validation of our findings.

The Impact of Stroma Admixture on Molecular Subtypes and Prognostic Gene Signatures in Serous Ovarian Cancer

BACKGROUND

Recent efforts to improve outcomes for high-grade serous ovarian cancer, a leading cause of cancer death in women, have focused on identifying molecular subtypes and prognostic gene signatures, but existing subtypes have poor cross-study robustness. We tested the contribution of cell admixture in published ovarian cancer molecular subtypes and prognostic gene signatures.

METHODS

Gene signatures of tumor and stroma were developed using paired microdissected tissue from two independent studies. Stromal genes were investigated in two molecular subtype classifications and 61 published gene signatures. Prognostic performance of gene signatures of stromal admixture was evaluated in 2,527 ovarian tumors (16 studies). Computational simulations of increasing stromal cell proportion were performed by mixing gene-expression profiles of paired microdissected ovarian tumor and stroma.

RESULTS

Recently described ovarian cancer molecular subtypes are strongly associated with the cell admixture. Tumors were classified as different molecular subtypes in simulations where the percentage of stromal cells increased. Stromal gene expression in bulk tumors was associated with overall survival (hazard ratio, 1.17; 95% confidence interval, 1.11-1.23), and in one data set, increased stroma was associated with anatomic sampling location. Five published prognostic gene signatures were no longer prognostic in a multivariate model that adjusted for stromal content.

CONCLUSIONS

Cell admixture affects the interpretation and reproduction of ovarian cancer molecular subtypes and gene signatures derived from bulk tissue. Elucidating the role of stroma in the tumor microenvironment and in prognosis is important.

IMPACT

Single-cell analyses may be required to refine the molecular subtypes of high-grade serous ovarian cancer.

Association of the tumour stroma percentage in the preoperative biopsies with lymph node metastasis in colorectal cancer

BACKGROUND

Preoperative prediction of lymph node (LN) status is integral to determining the most appropriate treatment strategy for colorectal cancer (CRC). This study aimed to develop and validate a nomogram to predict LN metastasis in CRC preoperatively.

METHODS

A total of 530 patients were enrolled and divided into training and validation cohorts. The tumour stroma percentage (TSP) of the preoperative biopsies was assessed. The risk factors for LN metastasis were selected, and a nomogram was constructed subsequently. The performance of the nomogram was assessed by using the AUROC and the calibration curve, and then validated in the validation cohort.

RESULTS

High TSP was significantly associated with LN metastasis in both the training and validation cohorts. Computed tomography (CT)-reported T stage, CT-reported LN status, preoperative tumour differentiation, carcinoembryonic antigen, carbohydrate antigen 19-9 and TSP were independent predictors of LN metastasis in CRC. A nomogram incorporating the six predictors was constructed. The nomogram yielded good discrimination and calibration, with an AUROC of 0.846 (95% CI: 0.807-0.886) and 0.809 (95% CI: 0.745-0.872) in the training and validation cohorts, respectively.

CONCLUSIONS

Assessment of TSP in the preoperative biopsies provided additional information about the LN status. The nomogram was useful for tailored therapy in CRC preoperatively.

Whole-Slide Image Analysis Reveals Quantitative Landscape of Tumor-Immune Microenvironment in Colorectal Cancers

PURPOSE

Despite the well-known prognostic value of the tumor-immune microenvironment (TIME) in colorectal cancers, objective and readily applicable methods for quantifying tumor-infiltrating lymphocytes (TIL) and the tumor-stroma ratio (TSR) are not yet available.

EXPERIMENTAL DESIGN

We established an open-source software-based analytic pipeline for quantifying TILs and the TSR from whole-slide images obtained after CD3 and CD8 IHC staining. Using a random forest classifier, the method separately quantified intraepithelial TILs (iTIL) and stromal TILs (sTIL). We applied this method to discovery and validation cohorts of 578 and 283 stage III or high-risk stage II colorectal cancers patients, respectively, who were subjected to curative surgical resection and oxlaliplatin-based adjuvant chemotherapy.

RESULTS

Automatic quantification of iTILs and sTILs showed a moderate concordance with that obtained after visual inspection by a pathologist. The K-means-based consensus clustering of 197 TIME parameters that showed robustness against interobserver variations caused colorectal cancers to be grouped into five distinctive subgroups, reminiscent of those for consensus molecular subtypes (CMS1-4 and mixed/intermediate group). In accordance with the original CMS report, the CMS4-like subgroup (cluster 4) was significantly associated with a worse 5-year relapse-free survival and proved to be an independent prognostic factor. The clinicopathologic and prognostic features of the TIME subgroups have been validated in an independent validation cohort.

CONCLUSIONS

Machine-learning-based image analysis can be useful for extracting quantitative information about the TIME, using whole-slide histopathologic images. This information can classify colorectal cancers into clinicopathologically relevant subgroups without performing a molecular analysis of the tumors.

Targeting of the Cancer-Associated Fibroblast-T-Cell Axis in Solid Malignancies

The introduction of a wide range of immunotherapies in clinical practice has revolutionized the treatment of cancer in the last decade. The majority of these therapeutic modalities are centered on reinvigorating a tumor-reactive cytotoxic T-cell response. While impressive clinical successes are obtained, the majority of cancer patients still fail to show a clinical response, despite the fact that their tumors express antigens that can be recognized by the immune system. This is due to a series of other cellular actors, present in or attracted towards the tumor microenvironment, including regulatory T-cells, myeloid-derived suppressor cells and cancer-associated fibroblasts (CAFs). As the main cellular constituent of the tumor-associated stroma, CAFs form a heterogeneous group of cells which can drive cancer cell invasion but can also impair the migration and activation of T-cells through direct and indirect mechanisms. This singles CAFs out as an important next target for further optimization of T-cell based immunotherapies. Here, we review the recent literature on the role of CAFs in orchestrating T-cell activation and migration within the tumor microenvironment and discuss potential avenues for targeting the interactions between fibroblasts and T-cells.

Predictive Values of MRI and PET Derived Quantitative Parameters for Patterns of Failure in Both p16+ and p16- High Risk Head and Neck Cancer

Purpose: FDG-PET adds to clinical factors, such tumor stage and p16 status, in predicting local (LF), regional (RF), and distant failure (DF) in poor prognosis locally advanced head and neck cancer (HNC) treated with chemoradiation. We hypothesized that MRI-based quantitative imaging (QI) metrics could add to clinical predictors of treatment failure more significantly than FDG-PET metrics. Materials and methods: Fifty four patients with poor prognosis HNCs who were enrolled in an IRB approved prospective adaptive chemoradiotherapy trial were analyzed. MRI-derived gross tumor volume (GTV), blood volume (BV), and apparent diffusion coefficient (ADC) pre-treatment and mid-treatment (fraction 10), as well as pre-treatment FDG PET metrics, were analyzed in primary and individual nodal tumors. Cox proportional hazards models for prediction of LRF and DF free survival were used to test the additional value of QI metrics over dominant clinical predictors. Results: The mean ADC pre-RT and its change rate mid-treatment were significantly higher and lower in p16- than p16+ primary tumors, respectively. A Cox model identified that high mean ADC pre-RT had a high hazard for LF and RF in p16- but not p16+ tumors (p = 0.015). Most interesting, persisting subvolumes of low BV (TV_bv) in primary and nodal tumors mid-treatment had high-risk for DF (p < 0.05). Also, total nodal GTV mid-treatment, mean/max SUV of FDG in all nodal tumors, and total nodal TLG were predictive for DF (p < 0.05). When including clinical stage (T4/N3) and total nodal GTV in the model, all nodal PET parameters had a p-value of >0.3, and only TV_bv of primary tumors had a p-value of 0.06. Conclusion: MRI-defined biomarkers, especially persisting subvolumes of low BV, add predictive value to clinical variables and compare favorably with FDG-PET imaging markers. MRI could be well-integrated into the radiation therapy workflow for treatment planning, response assessment, and adaptive therapy.

Tumour-stroma ratio and 5-year mortality in gastric adenocarcinoma: a systematic review and meta-analysis

Tumour-stroma ratio (TSR) is a novel potential prognostic factor in cancers and based on the proportions of stroma and tumour area. The prognostic value of TSR in gastric cancer is incompletely known. The aim of this study was to estimate prognostic significance of TSR in gastric adenocarcinoma. A search of PubMed (MEDLINE), Web of Science, EMBASE, Cochrane and Scopus databases was performed. A meta-analysis was conducted on five-year survival in gastric cancer patients using inverse variance random-effects methods. The literature search yielded 5329 potential titles, of which a total of seven studies were eligible. Results of six studies including a total of 1779 patients were pooled in the meta-analysis. Only 23 (1.3%) of the patients received neoadjuvant therapy. All six studies had a cut-off of 50% for the proportion of stroma when dividing the patients into low- and high stroma groups. Low TSR (high amount of stroma) was strongly associated with increased five-year mortality (hazard ratio 2.19, 95% CI 1.69–2.85). In conclusion, TSR is a strong prognostic factor in gastric cancer. It could be used to estimate prognosis of gastric cancer patients not receiving neoadjuvant chemotherapy. Further studies including patients receiving neoadjuvant therapy are recommended.

Preoperative, biopsy-based assessment of the tumour microenvironment in patients with primary operable colorectal cancer

The tumour microenvironment (TME) is recognised as an important prognostic characteristic and therapeutic target in patients with colorectal cancer (CRC). However, assessment generally utilises surgically resected specimens, precluding neoadjuvant targeting. The present study investigated the feasibility of intra‐epithelial CD3⁺ T‐lymphocyte density and tumour stroma percentage (TSP) assessment using preoperative colonoscopic biopsies from 115 patients who had undergone resection of stages I–III CRC, examining the relationship between biopsy and surgically resected specimen‐based assessment, and the relationship with cancer‐specific survival (CSS). High biopsy CD3⁺ density was associated with high CD3⁺ density in the invasive margin, cancer stroma and intra‐epithelial compartments of surgically resected specimens (area under the curve > 0.62, p < 0.05 for all) and with high Immunoscore. High biopsy TSP predicted high TSP in resected specimens (p = 0.001). Intra‐class correlation coefficient for both measures was >0.7 (p < 0.001), indicating excellent concordance between individuals. Biopsy CD3⁺ density (hazard ratio [HR] 0.23, p = 0.002) and TSP (HR 2.23, p = 0.029) were independently associated with CSS; this was comparable to the prognostic value of full section assessment (HR 0.21, p = 0.004, and HR 2.25, p = 0.033 respectively). These results suggest that assessment of the TME is comparable in biopsy and surgically resected specimens from patients with CRC, and biopsy‐based assessment could allow for stratification prior to surgery or commencement of therapy targeting the TME.

Tumour Budding and Tumour Stroma Ratio are Reliable Predictors for Death and Recurrence in Elderly Stage I Colon Cancer Patients

AIM

Tumour budding (BD) and Tumour stroma ratio (TSR) are considered valuable survival parameters for colon cancer (CC), but it is still unclear whether these parameters predict a poor prognosis. This study aimed to determine the survival effect of TB and TSR in elderly stage I CC patients.

METHODS

We evaluated these parameters in eighty-eight stage I CC patients who underwent surgical treatment alone between 1998 and 2015. The relationship between these parameters and age was investigated. Model A was used for methodology.

RESULTS

In elderly patients (≥68), these parameters obtained more successful results for following analysis: relationship between prognostic factors [lymphatic invasion (BD, p = 0.003; TSR, p = 0.003), perineural invasion (BD, p = 0.016; TSR, p = 0.004), tumour deposites (BD, p = 0.005; TSR, p = 0.033), MSI (BD, p=0.031; TSR, p=0.012), etc.], correlation of estimates (BD, r=0.724; TSR, r=0.703), and reproducibility of study (BD [Kappa (ĸ)]=0.53-0.75; TSR (ĸ) =0.56-0.71). Also, the cut-off values were useful for BD (area of under ROC (AUC)=0.816 [0.707-0.925]) and TSR (AUC=0.810 [0.697-0.924]). In univariate analysis, in elderly patients, these parameters had a poor 5-year survival for RFS (p<0.001[BD], p=0.001[TSR]), OS (p<0.001[BD], p=0.005[TSR]), and LR (p=0.008[BD], p=0.034[TSR]). Multivariate analysis confirmed that these parameters are independent worse predictors for RFS (BD: 1.42[HR], p=0.002; TSR: 1.50[HR], p=0.001), OS (BD: 1.38[HR], p=0.014; TSR: 1.42[HR], p=0.005), and LR (BD: 1.61[HR], p=0.034) in elderly patients.

CONCLUSIONS

Our results confirm that BD and TSR are reliable indexes for poor survival in elderly stage I CC patients. We recommend using model A for successful results and standardization.

Prognostic value of microvessel density in stage II and III colon cancer patients: a retrospective cohort study

Background

Microvessel density (MVD), as a derived marker for angiogenesis, has been associated with poor outcome in several types of cancer. This study aimed to evaluate the prognostic value of MVD in stage II and III colon cancer and its relation to tumour-stroma-percentage (TSP) and expression of HIF1A and VEGFA.

Methods

Formalin-fixed paraffin-embedded (FFPE) colon cancer tissues were collected from 53 stage II and 54 (5-fluorouracil-treated) stage III patients. MVD was scored by digital morphometric analysis of CD31-stained whole tumour sections. TSP was scored using haematoxylin-eosin stained slides. Protein expression of HIF1A and VEGFA was determined by immunohistochemical evaluation of tissue microarrays.

Results

Median MVD was higher in stage III compared to stage II colon cancers (11.1% versus 5.6% CD31-positive tissue area, p < 0.001). High MVD in stage II patients tended to be associated with poor disease free survival (DFS) in univariate analysis (p = 0.056). In contrast, high MVD in 5FU-treated stage III patients was associated with better DFS (p = 0.006). Prognostic value for MVD was observed in multivariate analyses for both cancer stages.

Conclusions

MVD is an independent prognostic factor associated with poor DFS in stage II colon cancer patients, and with better DFS in stage III colon cancer patients treated with adjuvant chemotherapy.

Electronic supplementary material

The online version of this article (10.1186/s12876-019-1063-4) contains supplementary material, which is available to authorized users.

Uniform Noting for International Application of the Tumor-Stroma Ratio as an Easy Diagnostic Tool: Protocol for a Multicenter Prospective Cohort Study

Background

Colon cancer treatment is dependent on the stage at diagnosis. The current Tumor-Node-Metastasis (TNM) staging for the selection of patients for adjuvant chemotherapy needs additional prognostic and predictive biomarkers. Better decision making for chemotherapy will result in reducing over- and undertreatment. We developed a new, easy-to-apply, practice-changing method to select colon cancer patients for adjuvant chemotherapy: the tumor-stroma ratio (TSR). The TSR distinguishes within stage II-III patients who will likely benefit from adjuvant chemotherapy and those who will not.

Objective

The aim of the study was to add, in addition to the TNM classification, the TSR to current routine pathology evaluation. Pathologists will be instructed for scoring the TSR in combination with a quality assessment program. An international multicenter study will validate the parameter prospectively.

Methods

The study is designed for future implementation of the TSR to the current TNM guidelines, using routinely Haematoxylin- and Eosin-stained tumor tissue sections. In part 1 of the study, an electronic learning (e-learning) module with a quality assessment program using the European Society of Pathology framework will be developed. This module will be used to assess the reliability and reproducibility of the TSR, conducted by national and international pathologists. Part 2 will involve the validation of the TSR in a prospective cohort of colon cancer p-stage II-III patients in a multicenter setting. In total, 1500 patients will be included.

Results

The results of part 1 will be expected in the first half of 2019. For part 2, the inclusion of patients in the prospective study, which started at the end of 2018, will take 3 years with an additional follow-up after another 3 years.

Conclusions

The main endpoints of this study are as follows: in part 1, trained (international) pathologists who are able to reliably score the TSR, resulting in low intra- and interobserver variation; in part 2, confirmation of significant survival differences for patients with a stroma-high tumor versus patients with a stroma-low tumor. On the basis of these findings, a modification in current treatment guidelines will be suggested.

Trial Registration

Netherlands Trial Register NTR7270; https://www.trialregister.nl/trial/7072

International Registered Report Identifier (IRRID)

DERR1-10.2196/13464

Stromal organization as predictive biomarker for the treatment of colon cancer with adjuvant bevacizumab; a post-hoc analysis of the AVANT trial

PURPOSE

Intra-tumoral stroma has become increasingly important in understanding tumor biology, tumor progression and clinical outcome. The amount itself, quantified as the tumor-stroma ratio (TSR), has proven to be prognostic in stage I-III colon cancer. Also, alterations in stromal organization have been found to provide prognostic and predictive information in certain cancers. Here, we evaluated the predictive value of stromal organization in high-risk stage II and III colon cancer with respect to adjuvant bevacizumab and chemotherapy.

METHODS

In a post-hoc analysis, stromal organization was microscopically determined in hematoxylin and eosin-stained primary tumor tissue samples of 1226 patients enrolled in the AVANT trial.

RESULTS

We found that patients with tumors with a disorganized stroma showed different survival rates after the addition of bevacizumab compared to standard oxaliplatin-based chemotherapy regimens. However, overall this difference was not significant with a HR of 0.94 (95% CI 0.57-1.55; p = 0.80) for disease-free survival (DFS) and 1.01 (95% CI 0.51-1.99; p = 0.99) for overall survival (OS). Subgroup analysis, however, revealed that stromal organization combined with TSR allowed the identification of stroma-high patients with absolute cumulative survival benefits up to 15% when bevacizumab was added to oxaliplatin-based chemotherapy regimens.

CONCLUSIONS

In high-risk stage II and stage III colon cancer, we found that subgroup analysis of the combined parameters stromal organization and TSR allows for the identification of patients with absolute cumulative DFS and OS benefits of up to 15%, when adding bevacizumab to the currently recommended oxaliplatin-based chemotherapy. Stromal organization itself does, however, not serve as an independent prognostic or predictive parameter.

Prognostic significance of inflammatory cell response in patients with colorectal cancer

Cancer cells are characterized by a low antigenic immunogenicity, a rapid growth and an immunosuppressive effect on the extracellular matrix. These properties induce a weak immune response in colorectal cancer (CRC) carcinogenesis. It is therefore crucial to determine the composition of the inflammatory mass, including neutrophils, macrophages and eosinophils in the tumor tissue of patients with CRC, and to analyze other clinicopathological parameters. The present study included 144 patients diagnosed with CRC. Tissue samples obtained from routine histopathological diagnosis were stained with hematoxylin and eosin. Inflammatory cells were assessed in the invasive front and in the center of the tumor by light microscopy under a high-power magnification. The percentage of neutrophils in the invasive front was significantly higher compared with that in the center of the tumor mass (P<0.01). Macrophages and eosinophils were present in the invasive front and in the center of tumor mass in most cases. The presence of neutrophils, macrophages and eosinophils was correlated with various clinicopathological features. Patients with macrophages present in the center of tumor mass had longer disease-free survival time (P=0.041). In conclusion, the present study demonstrated that the inflammatory cell infiltrate served a significant role in the immune response of patients with CRC. It should be noted that the presence of macrophages localized in the stroma of the central part of the primary tumor mass was associated with the survival time of patients with CRC.

Cancer-associated fibroblasts in gastrointestinal cancer

The tumour microenvironment, also termed the tumour stroma or tumour mesenchyme, includes fibroblasts, immune cells, blood vessels and the extracellular matrix and substantially influences the initiation, growth and dissemination of gastrointestinal cancer. Cancer-associated fibroblasts (CAFs) are one of the critical components of the tumour mesenchyme and not only provide physical support for epithelial cells but also are key functional regulators in cancer, promoting and retarding tumorigenesis in a context-dependent manner. In this Review, we outline the emerging understanding of gastrointestinal CAFs with a particular emphasis on their origin and heterogeneity, as well as their function in cancer cell proliferation, tumour immunity, angiogenesis, extracellular matrix remodelling and drug resistance. Moreover, we discuss the clinical implications of CAFs as biomarkers and potential targets for prevention and treatment of patients with gastrointestinal cancer.

Tumor-associated Macrophages as Prognostic and Predictive Biomarkers for Postoperative Adjuvant Chemotherapy in Patients with Stage II Colon Cancer

PURPOSE

For stage II colon cancer, the efficacy of postoperative adjuvant chemotherapy remains controversial. It is well known that tumor-associated macrophages (TAMs) are important in tumor progression. In this study, TAMs were investigated as prognostic and predictive biomarkers for the efficacy of adjuvant chemotherapy for stage II colon cancer after radical resection.

EXPERIMENTAL DESIGN

This study enrolled two independent cohorts of consecutive patients from one medical center with pathologic stage II colon cancer after radical resections. Macrophages were detected using IHC staining of CD68 and CD206. Infiltration densities of CD68⁺ TAMs, CD206⁺ TAMs, and ratio of CD206⁺ TAMs/CD68⁺ TAMs (CD206/CD68 ratio) were calculated as prognostic and predictive biomarkers.

RESULTS

The primary and validation cohorts consisted of 521 and 314 patients, respectively. In both cohorts, high CD206/CD68 ratio was significantly associated with poor disease-free survival (DFS) and overall survival (OS). As an independent risk factor, CD206/CD68 ratio also had significantly better prognostic efficacy than CD68⁺ TAM density, CD206⁺ TAM density, and traditional clinicopathologic high-risk factors. Moreover, adjuvant chemotherapy significantly improved DFS and OS for patients with high CD206/CD68 ratio but not for those with low CD206/CD68 ratio. The interaction analyses were also significant for DFS. In subgroup analysis, CD206/CD68 ratio was still a significant predictor for adjuvant chemotherapy for patients in traditional high-risk group of recurrence (significant interaction for DFS).

CONCLUSIONS

For stage II colon cancer, CD206/CD68 ratio is a better prognostic and predictive biomarker for postoperative adjuvant chemotherapy. Together with clinicopathologic high-risk factors, it will aid in precision treatment.

Increased expression of cancer-associated fibroblast markers at the invasive front and its association with tumor-stroma ratio in colorectal cancer

Background

The tumor microenvironment has a critical role in regulating cancer cell behavior. Tumors with high stromal content are associated with poor patient outcome. The tumor-stroma ratio (TSR) identifies colorectal cancers (CRC) with poor patient prognosis based on hematoxylin & eosin stained sections. The desmoplastic reaction consists to a great extent of cancer-associated fibroblasts (CAFs) of which different subtypes are known. The aim of this study is to investigate and quantify CAFs present in the tumor stroma of CRC stratified by the TSR to possibly add prognostic significance to the TSR.

Methods

The expression of established CAF markers was compared between stroma-low and stroma-high tumors using transcriptomic data of 71 stage I – III CRC. Based on literature, fibroblast and stromal markers were selected to perform multiplex immunofluorescent staining on formalin fixed, paraffin-embedded tumor sections of patients diagnosed with stage III colon cancer. Antibodies against the following markers were used: αSMA, PDGFR -β, FAP, FSP1 and the stromal markers CD45 and CD31 as reference. The markers were subsequently quantified in the stroma using the Vectra imaging microscope.

Results

The transcriptomic data showed that all CAF markers except one were higher expressed in stroma-high compared to stroma-low tumors. Histologically, stroma-high tumors showed a decreased number of FSP1⁺/CD45⁺ cells and a trend of an increased expression of FAP compared to stroma-low tumors. FAP was higher expressed at the invasive part compared to the tumor center in both stroma-high and stroma-low tumors.

Conclusions

The increased expression of FAP at the invasive part and in stroma-high tumors might contribute to the invasive behavior of cancer cells. Future functional experiments should investigate the contribution of FAP to cancer cell invasion. Combining the quantity of the stroma as defined by the TSR with the activity level of CAFs using the expression of FAP may result in an expanded stroma-based tool for patient stratification.

Electronic supplementary material

The online version of this article (10.1186/s12885-019-5462-2) contains supplementary material, which is available to authorized users.

Prognostic value of low CDX2 expression in colorectal cancers with a high stromal content - a short report

PURPOSE

Lack of expression of the intestinal transcription factor CDX2 in colorectal cancer (CRC) identifies patients with a poor prognosis. This biomarker has previously been suggested to be prognostic in CRCs with a high stromal content based on mRNA expression data. We investigated the prognostic value of CDX2 expression in microsatellite stable CRC stratified by stromal content using microscopy-based techniques.

METHODS AND RESULTS

The study included a cohort of 236 patients with stage I-IV CRC. We assessed by microscopy the tumour-stroma ratio (TSR) and the immunohistochemical CDX2 intensity. We found that patients of the stroma-high group had a worse prognosis compared to those of the stroma-low group [disease-free survival in a multivariate analysis (DFS_multivariate) HR 1.52 (95% CI 1.05-2.21)]. In our cohort, low CDX2 expression (14.6%) showed prognostic value for DFS_multivariate [HR 1.93 (95% CI 1.16-3.23)]. Interestingly, when stratifying the cohort by TSR, no prognostic difference was observed related to CDX2 expression in stroma-low tumours. However, CDX2 expression was found to be prognostic within the stroma-high group [DFS_multivariate HR 3.02 (95% CI 1.49-6.13)]. The p value for interaction between TSR and CDX2 status was borderline significant in DFS (p = 0.071).

CONCLUSIONS

The present study confirms a poor outcome of patients with stroma-high tumours. Low CDX2 expression in tumours with a high stromal content identified patients with a particularly poor prognosis. The present study did not reveal a clear difference in TSR associated with CDX2 status and survival. This method, solely based on microscopy, identifies patients who have a high risk of relapse and a poor outcome, and who may benefit from targeted therapy.

Computer aided quantification of intratumoral stroma yields an independent prognosticator in rectal cancer

PURPOSE

Tumor-stroma ratio (TSR) serves as an independent prognostic factor in colorectal cancer and other solid malignancies. The recent introduction of digital pathology in routine tissue diagnostics holds opportunities for automated TSR analysis. We investigated the potential of computer-aided quantification of intratumoral stroma in rectal cancer whole-slide images.

METHODS

Histological slides from 129 rectal adenocarcinoma patients were analyzed by two experts who selected a suitable stroma hot-spot and visually assessed TSR. A semi-automatic method based on deep learning was trained to segment all relevant tissue types in rectal cancer histology and subsequently applied to the hot-spots provided by the experts. Patients were assigned to a 'stroma-high' or 'stroma-low' group by both TSR methods (visual and automated). This allowed for prognostic comparison between the two methods in terms of disease-specific and disease-free survival times.

RESULTS

With stroma-low as baseline, automated TSR was found to be prognostic independent of age, gender, pT-stage, lymph node status, tumor grade, and whether adjuvant therapy was given, both for disease-specific survival (hazard ratio = 2.48 (95% confidence interval 1.29-4.78)) and for disease-free survival (hazard ratio = 2.05 (95% confidence interval 1.11-3.78)). Visually assessed TSR did not serve as an independent prognostic factor in multivariate analysis.

CONCLUSIONS

This work shows that TSR is an independent prognosticator in rectal cancer when assessed automatically in user-provided stroma hot-spots. The deep learning-based technology presented here may be a significant aid to pathologists in routine diagnostics.

Predicting survival from colorectal cancer histology slides using deep learning: A retrospective multicenter study

BACKGROUND

For virtually every patient with colorectal cancer (CRC), hematoxylin-eosin (HE)-stained tissue slides are available. These images contain quantitative information, which is not routinely used to objectively extract prognostic biomarkers. In the present study, we investigated whether deep convolutional neural networks (CNNs) can extract prognosticators directly from these widely available images.

METHODS AND FINDINGS

We hand-delineated single-tissue regions in 86 CRC tissue slides, yielding more than 100,000 HE image patches, and used these to train a CNN by transfer learning, reaching a nine-class accuracy of >94% in an independent data set of 7,180 images from 25 CRC patients. With this tool, we performed automated tissue decomposition of representative multitissue HE images from 862 HE slides in 500 stage I-IV CRC patients in the The Cancer Genome Atlas (TCGA) cohort, a large international multicenter collection of CRC tissue. Based on the output neuron activations in the CNN, we calculated a "deep stroma score," which was an independent prognostic factor for overall survival (OS) in a multivariable Cox proportional hazard model (hazard ratio [HR] with 95% confidence interval [CI]: 1.99 [1.27-3.12], p = 0.0028), while in the same cohort, manual quantification of stromal areas and a gene expression signature of cancer-associated fibroblasts (CAFs) were only prognostic in specific tumor stages. We validated these findings in an independent cohort of 409 stage I-IV CRC patients from the "Darmkrebs: Chancen der Verhütung durch Screening" (DACHS) study who were recruited between 2003 and 2007 in multiple institutions in Germany. Again, the score was an independent prognostic factor for OS (HR 1.63 [1.14-2.33], p = 0.008), CRC-specific OS (HR 2.29 [1.5-3.48], p = 0.0004), and relapse-free survival (RFS; HR 1.92 [1.34-2.76], p = 0.0004). A prospective validation is required before this biomarker can be implemented in clinical workflows.

CONCLUSIONS

In our retrospective study, we show that a CNN can assess the human tumor microenvironment and predict prognosis directly from histopathological images.

Histopathological and molecular classification of colorectal cancer and corresponding peritoneal metastases

BACKGROUND

Patients with colorectal peritoneal carcinomatosis have a very poor prognosis. The recently developed consensus molecular subtype (CMS) classification of primary colorectal cancer categorizes tumours into four robust subtypes, which could guide subtype-targeted therapy. CMS4, also known as the mesenchymal subtype, has the greatest propensity to form distant metastases. CMS4 status and histopathological features of colorectal peritoneal carcinomatosis were investigated in this study.

METHODS

Fresh-frozen tissue samples from primary colorectal cancer and paired peritoneal metastases from patients who underwent cytoreductive surgery combined with hyperthermic intraperitoneal chemotherapy were collected. Histopathological features were analysed, and a reverse transcriptase-quantitative PCR test was used to assess CMS4 status of all collected lesions.

RESULTS

Colorectal peritoneal carcinomatosis was associated with adverse histopathological characteristics, including a high percentage of stroma in both primary tumours and metastases, and poor differentiation grade and high-grade tumour budding in primary tumours. Furthermore, CMS4 was significantly enriched in primary tumours with peritoneal metastases, compared with unselected stage I-IV tumours (60 per cent (12 of 20) versus 23 per cent; P = 0.002). The majority of peritoneal metastases (75 per cent, 21 of 28) were also classified as CMS4. Considerable intrapatient subtype heterogeneity was observed. Notably, 15 of 16 patients with paired tumours had at least one CMS4-positive tumour location.

CONCLUSION

Significant enrichment for CMS4 was observed in colorectal peritoneal carcinomatosis. Surgical relevance Cytoreductive surgery combined with hyperthermic intraperitoneal chemotherapy (CRS-HIPEC) improves survival of selected patients with colorectal peritoneal carcinomatosis, but recurrence is common. Histopathological and molecular analysis of colorectal peritoneal carcinomatosis could provide clues for development of novel therapies. In this study, colorectal peritoneal carcinomatosis was found to be enriched for tumours with high stromal content and CMS4-positive status. To further improve prognosis for patients with colorectal peritoneal carcinomatosis, therapies that target tumour-stroma interaction could be added to CRS-HIPEC.

Obesity and gastrointestinal cancer: the interrelationship of adipose and tumour microenvironments

Increasing recognition of an association between obesity and many cancer types exists, but how the myriad of local and systemic effects of obesity affect key cellular and non-cellular processes within the tumour microenvironment (TME) relevant to carcinogenesis, tumour progression and response to therapies remains poorly understood. The TME is a complex cellular environment in which the tumour exists along with blood vessels, immune cells, fibroblasts, bone marrow-derived inflammatory cells, signalling molecules and the extracellular matrix. Obesity, in particular visceral obesity, might fuel the dysregulation of key pathways relevant to both the adipose microenvironment and the TME, which interact to promote carcinogenesis in at-risk epithelium. The tumour-promoting effects of obesity can occur at the local level as well as systemically via circulating inflammatory, growth factor and metabolic mediators associated with adipose tissue inflammation, as well as paracrine and autocrine effects. This Review explores key pathways linking visceral obesity and gastrointestinal cancer, including inflammation, hypoxia, altered stromal and immune cell function, energy metabolism and angiogenesis.

The prognostic value of tumor-stroma ratio in tumor-positive axillary lymph nodes of breast cancer patients

The tumor-stroma ratio (TSR) has previously been found to be a strong prognostic parameter in primary breast cancer tumors. Since the presence of tumor cells in lymph nodes is important for clinical decision making, the influence of TSR in the primary breast tumor combined with the TSR in tumor-positive lymph nodes on prognosis was evaluated. Women with invasive breast cancer without distant metastasis who underwent an axillary lymph node dissection between 1985 and 1994 at the Leiden University Medical Center were retrospectively analyzed. TSR assessment was performed on hematoxylin and eosin stained tissue slides. In total, 87 (45.5%) primary tumors were scored as stroma-low and 104 (54.5%) as stroma-high. Patients with a high stromal percentage in the primary tumors had a statistically significant worse relapse free period (RFP) compared to stroma-low tumors (HR 1.97, 95% CI 1.37-2.82, p < 0.001). A total number of 915 lymph nodes were assessed for TSR. In 101 (52.9%) patients, heterogeneity was observed between stroma percentage category in primary tumor and lymph nodes. The combination of TSR of the primary tumor combined with TSR of tumor-positive lymph nodes strengthened each other as independent prognostic parameter for RFP (p = 0.019). Patients with primary tumor stroma-low/lymph nodes stroma-low tumors showed strongly improved RFP rates compared to patients with primary tumor stroma-high/lymph node stroma-high tumors with 10-year percentages of 58 versus 8%, respectively. Assessing the TSR on tumor-positive lymph nodes can provide additional prognostic information. Stromal activation strongly differs between primary tumors and lymph node metastasis.

The prognostic value of tumour-stroma ratio in primary breast cancer with special attention to triple-negative tumours: a review

Purpose

There is a strong need to improve the prognostication of breast cancer patients in order to prevent over- and undertreatment, especially when considering adjuvant chemotherapy. Tumour stroma characteristics might be valuable in predicting disease progression.

Methods

Studies regarding the prognostic value of tumour–stroma ratio (TSR) in breast cancer are evaluated.

Results

A high stromal content is related to a relatively poor prognosis. The most pronounced prognostic effect of this parameter seems to be observed in the triple-negative breast cancer (TNBC) subtype.

Conclusions

TSR assessment might represent a simple, fast and reproducible prognostic factor at no extra costs, and could possibly be incorporated into routine pathological diagnostics. Despite these advantages, a robust clinical validation of this parameter has yet to be established in prospective studies.

Electronic supplementary material

The online version of this article (10.1007/s10549-018-4987-4) contains supplementary material, which is available to authorized users.

Low tumor purity is associated with poor prognosis, heavy mutation burden, and intense immune phenotype in colon cancer

Purpose

Tumor purity is defined as the proportion of cancer cells in the tumor tissue. The impact of tumor purity on colon cancer (CC) prognosis, genetic profile, and microenvironment has not been thoroughly accessed.

Materials and methods

Clinical and transcriptomic data from three public datasets, GSE17536/17537, GSE39582, and TCGA, were retrospectively collected (n=1,248). Tumor purity of each sample was inferred by a computational method based on transcriptomic data. Survival-related analyses were performed on microarray dataset containing GSE17536/17537 and GSE39582 (n=794), whereas TCGA dataset was utilized for subsequent genomic analysis (n=454).

Results

Right-sided CC patients showed a significantly lower tumor purity. Low purity CC conferred worse survival, and tumor purity was identified as an independent prognostic factor. Moreover, high tumor purity CC patients benefited more from adjuvant chemotherapy. Subsequent genomic analysis found that the mutation burden was negatively associated with tumor purity, with only APC and KRAS significantly more mutated in high purity CC. However, no somatic copy number alteration event was correlated with tumor purity. Furthermore, immune-related pathways and immunotherapy-associated markers (programmed cell death protein 1 [PD-1], programmed death-ligand 1 [PD-L1], cytotoxic T-lymphocyte-associated protein 4 [CTLA-4], Lymphocyte-activation gene 3 [LAG-3] and T-cell immunoglobulin and mucin-domain containing-3 [TIM-3]) were highly enriched in low purity samples. Notably, the relative proportion of M2 macrophages and neutrophils, which indicated worse survival in CC, was negatively associated with tumor purity.

Conclusion

Tumor purity exhibited potential value for CC prognostic stratification as well as adjuvant chemotherapy benefit prediction. The relative worse survival in low purity CC may attribute to higher mutation frequency in key pathways and purity-related microenvironmental changing.

Novel digital signatures of tissue phenotypes for predicting distant metastasis in colorectal cancer

Distant metastasis is the major cause of death in colorectal cancer (CRC). Patients at high risk of developing distant metastasis could benefit from appropriate adjuvant and follow-up treatments if stratified accurately at an early stage of the disease. Studies have increasingly recognized the role of diverse cellular components within the tumor microenvironment in the development and progression of CRC tumors. In this paper, we show that automated analysis of digitized images from locally advanced colorectal cancer tissue slides can provide estimate of risk of distant metastasis on the basis of novel tissue phenotypic signatures of the tumor microenvironment. Specifically, we determine what cell types are found in the vicinity of other cell types, and in what numbers, rather than concentrating exclusively on the cancerous cells. We then extract novel tissue phenotypic signatures using statistical measurements about tissue composition. Such signatures can underpin clinical decisions about the advisability of various types of adjuvant therapy.

New Development of Biomarkers for Gastrointestinal Cancers: From Neoplastic Cells to Tumor Microenvironment

Biomarkers refer to a plethora of biological characteristics that can be quantified to facilitate cancer diagnosis, forecast the prognosis of disease, and predict a response to treatment. The identification of objective biomarkers is among the most crucial steps in the realization of individualized cancer care. Several tumor biomarkers for gastrointestinal malignancies have been applied in the clinical setting to help differentiate between cancer and other conditions, facilitate patient selection for targeted therapies, and to monitor treatment response and recurrence. With the coming of the immunotherapy age, the need for a new development of biomarkers that are indicative of the immune response to tumors are unprecedentedly urgent. Biomarkers from the tumor microenvironment, tumor genome, and signatures from liquid biopsies have been explored, but the majority have shown a limited prognostic or predictive value as single biomarkers. Nevertheless, use of multiplex biomarkers has the potential to provide a significantly increased diagnostic accuracy compared to traditional single biomarker. A comprehensive analysis of immune-biomarkers is needed to reveal the dynamic and multifaceted anti-tumor immunity and thus imply for the rational design of assays and combinational strategies.

Molecular profiling of colorectal tumors stratified by the histological tumor-stroma ratio - Increased expression of galectin-1 in tumors with high stromal content

The tumor microenvironment is a dominant determinant of cancer cell behavior. Reactive tumor stroma is associated with poor outcome perspective. The tumor-stroma ratio (TSR) is a strong independent prognostic factor in colorectal cancer and is easily assessed using conventional hematoxylin and eosin (H&E) stained paraffin sections at the invasive margin of the tumor. We aim to understand the biology of the tumor stroma in colorectal cancer by investigating the transcriptomic profiles of tumors classified by the TSR method. The TSR was assessed in a cohort of 71 colorectal cancer patients undergoing surgery without (neo)adjuvant therapy. In the cohort, stroma-high tumors were distinguished from stroma-low tumors at gene expression level in the upregulation of biological pathways related to extracellular matrix (ECM) remodeling and myogenesis. The activated microenvironment in stroma-high tumors overexpressed different types of collagen genes, THBS2 and 4 as well as INHBA, COX71A and LGALS1/galectin-1. The upregulation of THBS2, COX7A1 and LGALS1/galectin-1. The upregulation of THBS2, COX7A1 and LGALS1/galectin-1 in stroma-high tumors was validated in The Cancer Genome Atlas. In conclusion, the gene expression data reflects the high stromal content of tumors assessed based on the histological method, the TSR. The composition of the microenvironment suggests an altered proteolysis resulting in ECM remodeling and invasive capacity of tumor cells.

Scoring the tumor-stroma ratio in colon cancer: procedure and recommendations

The tumor-stroma ratio (TSR) has been reported as a strong, independent prognostic parameter in colon cancer as well as in other epithelial cancer types, and may be implemented to routine pathology diagnostics. The TSR is an easy technique, based on routine hematoxylin and eosin stained histological sections, estimating the amount of stroma present in the primary tumor. It links tumors with high stromal content to poor prognosis. The analysis time is less than 2 min with a low inter-observer variation. Scoring of the TSR has been validated in a number of independent international studies. In this manuscript, we provide a detailed technical description of estimating the TSR in colon cancer, including examples, pitfalls, and recommendations.

The cancer matrisome: From comprehensive characterization to biomarker discovery

Tumor progression and dissemination critically depend on support from the tumor microenvironment, the ensemble of cellular and acellular components surrounding and interacting with tumor cells. The extracellular matrix (ECM), the complex scaffolding of hundreds of proteins organizing cells in tissues, is a major component of the tumor microenvironment. It orchestrates cellular processes including proliferation, migration, and invasion, that are highly dysregulated during cancer progression. Alterations in ECM abundance, integrity, and mechanical properties have been correlated with poorer prognosis for cancer patients. Yet the ECM proteome, or "matrisome," of tumors remained until recently largely unexplored. This review will present the recent developments in computational and proteomic technologies that have allowed the comprehensive characterization of the ECM of different tumor types and microenvironmental niches. These approaches have resulted in the definition of protein signatures distinguishing tumors from normal tissues, tumors of different stages, primary from secondary tumors, and tumors from other diseased states such as fibrosis. Moreover, recent studies have demonstrated that the levels of expression of certain genes encoding ECM and ECM-associated proteins is prognostic of cancer patient survival and can thus serve as biomarkers. Last, proteomic studies have permitted the identification of novel ECM proteins playing functional roles in cancer progression. Such proteins have the potential to be exploited as therapeutic targets.

Prediction of pathological response to neoadjuvant treatment in rectal cancer with a two-protein immunohistochemical score derived from stromal gene-profiling

Background

Preoperative chemoradiotherapy followed by surgical mesorectal resection is the standard of care for locally advanced rectal carcinomas. Yet, predicting that patients will respond to treatment remains an unmet clinical challenge.

Experimental design

Using laser-capture microdissection we isolated RNA from stroma and tumour glands from prospective pre-treatment samples (n = 15). Transcriptomic profiles were obtained hybridising PrimeView Affymetrix arrays. We modelled a carcinoma-associated fibroblast-specific genes filtering data using GSE39396.

Results

The analysis of differentially expressed genes of stroma/tumour glands from responder and non-responder patients shows that most changes were associated with the stromal compartment; codifying mainly for extracellular matrix and ribosomal components. We built a carcinoma-associated fibroblast (CAF) specific classifier with genes showing changes in expression according to the tumour regression grade (FN1, COL3A1, COL1A1, MMP2 and IGFBP5). We assessed these five genes at the protein level by means of immunohistochemical staining in a patient's cohort (n = 38). For predictive purposes we used a leave-one-out cross-validated model with a positive predictive value (PPV) of 83.3%. Random Forest identified FN1 and COL3A1 as the best predictors. Rebuilding the leave-one-out cross-validated regression model improved the classification performance with a PPV of 93.3%. An independent cohort was used for classifier validation (n = 36), achieving a PPV of 88.2%. In a multivariate analysis, the two-protein classifier proved to be the only independent predictor of response.

Conclusion

We developed a two-protein immunohistochemical classifier that performs well at predicting the non-response to neoadjuvant treatment in rectal cancer.

Prognostic impact of tumor-stroma ratio in oral squamous cell carcinoma - A pilot study

BACKGROUND

Several prognostic indicators have been used for many decades in an attempt to predict clinical behaviour of Oral Squamous Cell Carcinoma (OSCC). The prognostic value of TSR is yet to be explored in OSCC. Hence, the aim of the present study was to evaluate the prognostic value of TSR in OSCC patients.

METHODOLOGY

A cohort of 60 histologically diagnosed cases of OSCC who underwent Radical Neck Dissection was included in the study. TSR was assessed and patients with >50% intratumor stroma were quantified as the stroma-poor group and those with <50% as the stroma-rich group.

RESULTS

The parametric tests were performed for the statistical evaluation of TSR with the clinico-pathological variables and the survival. The 3-year overall survival (OS) and disease-free survival (DFS) rates were 95.23% and 69.04%, respectively, in stroma-poor group and 77% and 44%, respectively in the stroma-rich group.

CONCLUSION

TSR may serve as a reliable histologic prognostic indicator in OSCC and could be used in routine diagnostic pathology.

Predictive potential of tumour-stroma ratio on benefit from adjuvant bevacizumab in high-risk stage II and stage III colon cancer

Background

The tumour–stroma ratio (TSR) has proven to be an independent prognostic factor in colon cancer.

methods

Haematoxylin eosin tissue slides of patients from the AVANT trial were microscopically scored for TSR and categorised as stroma -low or stroma -high. Scores were correlated to the primary and secondary endpoint disease-free survival (DFS) and overall survival (OS).

Results

Patients with stroma-high tumours (N = 339, 28%) had a significantly shorter DFS (p < 0.001) compared to stroma-low tumours (N = 824, 68%). In the bevacizumab-FOLFOX-4 arm, DFS was significantly shorter compared to FOLFOX-4 in stroma-low tumours, with a hazard ratio (HR) of 1.94 (95% CI 1.24–3.04; p = 0.004). In stroma-high tumours a trend for better DFS was seen in bevacizumab-FOLFOX-4 vs. FOLFOX-4 (HR 0.61 (95% CI 0.35–1.07; p = 0.08)). For bevacizumab-XELOX vs. FOLFOX-4, this was not seen (stroma-low HR 1.07 (95% CI 0.64–1.77; p = 0.80); stroma-high HR 0.78 (95% CI 0.47–1.30; p = 0.35)). OS showed the same pattern for bevacizumab-FOLFOX-4 vs. FOLFOX-4 with a HR of 2.53 (95% CI 1.36–4.71; p = 0.003) for stroma-low and HR 0.50 (95% CI 0.22–1.14; p = 0.10) for stroma-high tumours. For bevacizumab-XELOX vs. FOLFOX-4, HR 1.13 (95% CI 0.55–2.31; p = 0.74) for stroma-low tumours and HR 0.74 (95% CI 0.37–1.51; p = 0.41) for stroma-high tumours.

Conclusions

This exploratory analysis suggests a significantly shorter DFS and OS in stroma-low tumours with addition of bevacizumab to intravenous oxaliplatin-based chemotherapy, contrary to stroma-high tumours, where a beneficial trend is observed.

MiR-650 represses high-risk non-metastatic colorectal cancer progression via inhibition of AKT2/GSK3β/E-cadherin pathway

Although 5-year survival rate of non-metastatic colorectal cancer (CRC) is high, about 10% of patients in stage I and II still develop into metastatic CRC and eventually die after resection. Currently, there is no effective biomarker for predicting the prognosis of non-metastatic CRC in clinical practice. In this study, we identified miR-650 as a biomarker for prognosis prediction. We observed that the expression of miR-650 in tumor tissues had a positive association with overall survival. MiR-650 inhibited cell growth and invasion in vitro and in vivo. Furthermore, miR-650 targeted AKT2 and repressed the activation of the AKT pathway (AKT2/GSK3β/E-cadherin). Thus it induced the translocation of E-cadherin and β-catenin in cancer cells. Our results highlight the potential of miR-650 as a prognostic prediction biomarker and therapeutic target in non-metastatic CRC via inhibition of the AKT2/GSK3β/E-cadherin pathway.