Stromal scoring in advanced colon and rectal cancer: Stroma-rich tumors and their association with aggressive phenotypes

Received 2021-04-03 Received in revised form 2021-06-14 Accepted 2021-06-14 INTRODUCTION Cancer is currently recognized as a complex disease composed of several cell types, especially those derived from the surrounding mesenchymal stroma, with which neoplastic cells establish the tumor microenvironment (TME). In this environment, the tumor stroma represents one of the TME components (1–4). Tumor cells explore their stroma, changing its composition in a bi-directional communication, leading to the stromatogenesis. The interaction pathways are varied and complex. Therefore, the stromal tissue is not a passive component that involves the tumor (5). Studies have shown that tumor stroma plays a relevant and diverse role in tumorigenesis, acting in different stages: it facilitates the survival and proliferation of neoplastic cells; promotes epithelial-mesenchymal transition, and local and metastatic spread (6–12). Even in distant and lymph node metastatic sites, stromal components accompany cancer cells (13, 14). In malignant epithelial tumors, the scoring system based on the evaluation of the tumor-stroma proportion (TSP) in sections stained with hematoxylin and eosin (H&E) has been shown to be a good prognostic tool (9,10,15–18). Several international research groups have demonstrated that high amount of stroma contributes to a more aggressive tumor phenotypes (8, 9, 15–17, 19–21). Their goal was to fill the need for and identify new prognostic characteristics that could be used along with the current pathological staging (8, 20). The traditional tumor, lymph node and metastasis (TNM) system that has been used routinely for prognosis estimate and guidance of treatment for certain types of tumors (22, 23), lacks accuracy (21, 24, 25). In colorectal cancer (CRC), new reliable biomarkers are needed to guide personalized treatment (21) since current pathological variables only moderately indicate possible outcome and response to therapy (6, 19, 21). Currently, CRC represents a serious public health problem worldwide, occupying the third place in terms of incidence and the second place in mortality numbers (22), while being little attended by public policies in underdeveloped or developing countries (26). Although the complete biological role of stroma is not yet fully understood (20), in the last 10 years the evaluation of tumor stroma has gained interest due to its simplicity and, above all, to its clinical value as a potential prognostic factor. Furthermore, cancer cells and stroma are being considered as therapeutic targets in treatment strategies for solid tumors (14), in which the quantification of the stromal component may provide additional risk stratification for adaptation to neoadjuvant and adjuvant treatments (8, 15, 20, 27). The aim of this study was to evaluate the relevance of the tumor-stroma proportion and its association with confirmed aggressive phenotypes in a large series of patients diagnosed with cancer of the right/left colon and rectum in Brazil.


INTRODUCTION
Cancer is currently recognized as a complex disease composed of several cell types, especially those derived from the surrounding mesenchymal stroma, with which neoplastic cells establish the tumor microenvironment (TME). In this environment, the tumor stroma represents one of the TME components (1)(2)(3)(4). Tumor cells explore their stroma, changing its composition in a bi-directional communication, leading to the stromatogenesis. The interaction pathways are varied and complex. Therefore, the stromal tissue is not a passive component that involves the tumor (5). Studies have shown that tumor stroma plays a relevant and diverse role in tumorigenesis, acting in different stages: it facilitates the survival and proliferation of neoplastic cells; promotes epithelial-mesenchymal transition, and local and metastatic spread (6)(7)(8)(9)(10)(11)(12). Even in distant and lymph node metastatic sites, stromal components accompany cancer cells (13,14). In malignant epithelial tumors, the scoring system based on the evaluation of the tumor-stroma proportion (TSP) in sections stained with hematoxylin and eosin (H&E) has been shown to be a good prognostic tool (9,10,(15)(16)(17)(18). Several international research groups have demonstrated that high amount of stroma contributes to a more aggressive tumor phenotypes (8,9,(15)(16)(17)(19)(20)(21). Their goal was to fill the need for and identify new prognostic characteristics that could be used along with the current pathological staging (8,20). The traditional tumor, lymph node and metastasis (TNM) system that has been used routinely for prognosis estimate and guidance of treatment for certain types of tumors (22,23), lacks accuracy (21,24,25). In colorectal cancer (CRC), new reliable biomarkers are needed to guide personalized treatment (21) since current pathological variables only moderately indicate possible outcome and response to therapy (6,19,21). Currently, CRC represents a serious public health problem worldwide, occupying the third place in terms of incidence and the second place in mortality numbers (22), while being little attended by public policies in underdeveloped or developing countries (26). Although the complete biological role of stroma is not yet fully understood (20), in the last 10 years the evaluation of tumor stroma has gained interest due to its simplicity and, above all, to its clinical value as a potential prognostic factor. Furthermore, cancer cells and stroma are being considered as therapeutic targets in treatment strategies for solid tumors (14), in which the quantification of the stromal component may provide additional risk stratification for adaptation to neoadjuvant and adjuvant treatments (8,15,20,27). The aim of this study was to evaluate the relevance of the tumor-stroma proportion and its association with confirmed aggressive phenotypes in a large series of patients diagnosed with cancer of the right/left colon and rectum in Brazil.

METHODS
Tissue collection consisted of samples from colorectal adenocarcinoma (CRA) patients treated with curative surgery from year 2013 to 2018. Patients that received neoadjuvant treatment with radiotherapy and/or chemotherapy were excluded from the study. Histopathological data were obtained from the re-analysis of the histological slides of the surgical specimens and from the respective anatomical/ pathological report. Clinical data were obtained from the medical and hospital records. Evaluated clinicopathological data included age, gender, tumor topography, histological type, tumor grade (differentiation), depth of invasion, tumor budding, perineural invasion, angiolymphatic invasion, lymph node involvement, and presence of distant metastasis according to the Union Internationale Contre le Cancer / American Joint Cancer Committee (UICC/AJCC) TNM staging system (22,23). Tissue samples consisted of 4 µm thick, hematoxylin and eosin stained histological sections. The deepest invasive part of the bowel wall of the surgical specimen of primary tumor (slides used in routine pathology to determine pT-category) was used to evaluate the stromal proportion. Areas with the largest amount of stroma were selected by conventional microscopy using 2-5 × objective magnification. Subsequently, an area containing both tumor and stromal tissue was selected using 10 × objective magnification. Tumor cells were present at all edges of the selected image field (north-east-south-west) for quantification of the stromal component. Areas with large amount of muscle, mucus, necrosis or large vessels, as well as tears or tissue retraction artefacts were not included. Infiltration with inflammatory cells is not an exclusion criterion (28) and it was included in the scoring ( Figure 1). When there was more than one area with the high number of stroma visible, the area with the highest percentage of stroma was selected. In the case of only one doubtful area / field of view categorized as high stroma (even after consulting a second observer), the total composition of the entire tissue section was evaluated with the objective magnifications of 2 -to 5 × for classification of that particular case (28). Scoring percentages were semi-quantitatively reported as tenfold percents (10%, 20%, 30%, etc.). For statistical analyses and for the comparison with clinicopathological data samples were grouped as stroma-high (>50%, i.e. TSP low) and stroma-low (≤50%, i.e. TSP high) as previously described (4,9,10,(15)(16)(17). Statistical analyses were performed using SPSS V20 (SPSS Inc., Chicago, IL). Data were expressed as absolute and relative values, . A descriptive analysis of the variables was performed. Likelihood ratio test was used to investigate the effect of the variables on the stromal proportion. The level of significance was set at 0.05. The stromal scoring was estimated by two observers (RMSS and EMQ). The inter-observer agreement for TSP assessment, reported as categorical data, was determined using Kappa (κ) concordance index and the intra-class correlation coefficient (ICC). These criteria categorize a score of 0 as poor, 0 to 0.2 slight, 0.2 to 0.4 fair, 0.4 to 0.6 moderate, 0.6 to 0.8 substantial, and 0.8 to 1.0 almost perfect (29

RESULTS
Tissue collection Tissue collection consisted of 390 cases of CRA. The mean age was 63.5 years, with 81.8% of individuals over 50 years old and 6.4% of patients under 40 years old. The clinicopathological data were presented in Table 1.

DISCUSSION
Colorectal cancer (CRC) is increasing its incidence worldwide, particularly in low-income countries but also in developed countries andit already represents the second leading cause of death from cancer (22,37). Both morphological studies and recent molecular classifications of the CRC have highlighted the relevance of the tumor microenvironment, rekindling interest in the "seed and soil" hypothesis of colorectal carcinogenesis (4,38). Previous morphological studies have examined the stromal tumor relationships in the microenvironment of epithelial cancers (30)(31)(32)(33)(34)(35)(36). In our study tumor microenvironment and estimation of the proportion of tumor stroma in colorectal adenocarcinomas, in a middle-income country was evaluated.
In the surgically resected CRA stroma-high cases were detected in 53.3% of the patients, a substantially larger stromal proportion comparede to the average (24-29%) in initial studies (6-8, 19, 39), but similar to some more recent studies (5,13,(15)(16)(17) that pointed out a stromal estimate of up to 47%. The association found in this study between exclusively well-differentiated colorectal adenocarcinomas and low stroma was particular, and not seen in other series (5,8,15), indicating that the loss of differentiation was accompanied by stromal expansion. In a way to understand these findings, a pioneering study in the characterization and validation (in vitro and in vivo co-cultures) of the tumor microenvironment in CRC, observed that disordered extracellular matrix, with higher stroma proportion, drove a mesenchymal phenotype towards being similar to poorly differentiated (40). Our results showed that stroma-rich CRAs are significantly associated with adverse pathological characteristics, such as perineural invasion and lymphatic invasion, similarly to previous findings (15,20,41). Patients with a high stromal proportion more often had a tumor budding compared to stroma -poor tumors. The present findings may indicate a relevant role of tumor stroma in facilitating the differentiation, spread of tumor cells and in the epithelial-mensenchymal transition. This expressive association was also observed in other studies (8,17,20,42), although the complete elucidation of the molecular basis is still needed. A recent study showed that an increase in the composition of the tumor stroma and a reduction in epithelial cellularity were significantly correlated with genetic signatures related to epithelial-mensenchymal transition (12). The depth of invasion through the intestinal wall and the presence of lymph node metastases, which are recognized as the most important independent prognostic factors in CRC, related to survival and risk of recurrence, in addition to indicating the therapeutic approach (22,43,44), were significantly related to the percentage of stroma. All pT1 patients were stroma-low, while 75.6% of the pT2 patients were stroma-low. In the pT3 group, this percentage decreased to 41.4% and in the pT4 group it was 33.9%. Stroma-poor tumors showed less depth of invasion. These results point out that the expansion of the stromal compartment presents itself as a characteristic of aggressive disease, more locally advanced. The stromal proportion observed for the pT3-pT4 status was significantly higher, compared to early-stage tumors (TSP ≤50% -pT3: 58.6%; pT4: 66.1%; pT2: 24.4%; pT1: 0%; pTis: 0%). Percent of high stroma in patients with lymph node metastases was 67.6% (121 cases out of 179). The important association found between stromal estimation >50% and positive lymph nodes was also seen in other studies with colorectal cancer (8,15,20,45). The expansion of the stromal component was also a risk factor independent of the occurrence of distant metastases. Almost all stroma-rich CRC (80.6%) presented distant metastasis when surgicaly resected, an association also observed in a recent study (20). The observation of this association is particularly important in the therapeutic field, since the hypotheses indicate that the interruption of tumor-stroma interactions can inhibit or help eliminate tumor progression and metastasis (19), as well as that the stromal component may contribute to chemoresistance (8).
In recent years, a better understanding of the processes of carcinogenesis have fostered research into new biomarkers with the ultimate aim of promoting more personalized and effective therapy. In the case of CRC, especially for non-metastatic tumors (p-Stage II), which comprise a heterogeneous group with a different outcome (24) the classic histopathological classification is not sufficiantly informative for planning the treatment of these patients (21). The present study showed a strong association between stromal estimation and prognostic stage of the disease, that was in agreement with previous studies (6,7,45). Adenocarcinomas without lymph node metastasis or distant metastases had TSP ≤ 50%, while advanced disease (with lymph node and distant metastases) exhibited TSP > 50% thus demonstrating that stromal expansion in CRA represents a marker of tumor progression, associated with aggressive disease. No association was found between the tumor-stroma proportion and certain factors, in agreement with results from previous series, such as: male or female sex (5,7,8,13,41); age groups (8,13,41); location of the neoplasm (7,17,41) and histological type (13,15). The adoption of a morphometric methodological protocol widely used by independent international groups (9, 10, 15-18)for colon and rectal cancer, without new costs, proved to be highly reproducible, with good agreement by different observers confirming its applicability from previous studies (5-7, 9, 19, 46). It is important to highlight that the visual selection of the area analyzed by the pathologist is vital for the current approach in the CRA (47).

CONCLUSIONS
The description of the microscopic features, through hematoxylin and eosin staining, selected from the tumor microenvironment, alone or in combination, provides considerable information in colonic and rectal cancer. Tumor-stroma proportion is associated with all known histological characteristics, which reflect on an adverse clinical prognosis and guide the management of patients. The correlation between high stroma and advanced disease indicates that a high proportion of stromal tumor tissue can promote host tissue invasion and tumor aggressiveness, reflecting www.onk.ns.ac.rs/Archive • Published online June 2021 • https://doi.org/10.2298/AOO210403003S the defining contribution of the stromal compartment. The tumor-stroma proportion expresses the stage of the disease and can potentially complement the current staging system, even allowing a morphological correlation with the development of molecular classifications, in a precise, reproducible way and with applicability in routine diagnostics.