Epithelial malignant tumors initially invade the basement membrane and sequentially progress to the stroma. The interaction between the tumor cells and stroma, including the extracellular matrix, is important in the tumor progression of many cancers [1
]. On the other hand, tumor progression can be suppressed by peritumoral stromal components [2
]. The extracellular matrix of the stroma provides a dynamic network in tumor cells and affects tumor behavior [1
]. As the tumor size varies, the amount of stroma can also vary. The interface between tumor cells and stroma may be necessary for interpreting the role of the stroma. Colorectal cancer (CRC) is one of the common cancers worldwide [3
]. In normal colorectal tissue and CRC, lymphocytic infiltrations are frequently identified [4
]. In addition, the prognostic role of tumor-infiltrating lymphocytes and the immunoscore has been reported in CRC [7
]. These results suggest the crucial roles of intratumoral stroma in CRC. The tumor–stroma ratio (TSR) has been introduced as a simple parameter for evaluating the intratumoral stroma.
Although previous studies reported the prognostic impact of TSR in CRC, the results are controversial [9
]. Zengin et al. reported that tumors with a high proportion of stroma (>50% stroma, defined as stroma-high) were significantly correlated with worse survival, whereas tumors with an abundant carcinoma component (≤50% stroma, stroma-low) were associated with a better prognosis [14
]. In addition, there was no significant correlation between TSR and survival [13
]. A detailed analysis based on histologic subtypes has not been performed. Besides, the evaluation criteria for TSR are not fully elucidated. Basically, malignant tumors are evaluated for deciding therapy and predicting a patient’s prognosis through the tumor node metastatic (TNM) classification [15
]. Because TSR can easily be assessed through routine pathologic examinations, it is important to elucidate the evaluation criteria for TSR in CRC.
In the present study, we investigated the TSR of CRCs and evaluated the clinicopathological significance and prognostic implications of TSR. Also, we performed a subgroup analysis based on cutoffs of TSR and histologic subtypes. Angiogenesis is a multi-step process involving many angiogenesis-related molecules, such as hypoxia-inducible factor-1α (HIF-1α). Angiogenesis is regulated by multiple factors, many of which may individually predict microvessel density (MVD). Besides, since the tumor progression requires angiogenesis in the stroma, we compared the angiogenesis between stroma-low and stroma-high subgroups with the HIF-1α expression of tumor cells and MVD of the stroma.
The present study aimed to elucidate the clinicopathological significance and prognostic implications of TSR in CRC. In addition, the prognostic impact of TSR was determined according to evaluation criteria for stroma-low. We found that stroma-low was significantly correlated with less frequent vascular invasion and lower MVD than stroma-high. The present study is the first, to the best of our knowledge, to elucidate (1) the correlation between TSR and vascular invasion and MVD, and (2) the correlation between TSR and histologic subtypes in CRC.
TSR is defined as the proportion of the cancer cells relative to the surrounding stroma of the tumor [16
]. In the tumor microenvironment concept, intratumoral stroma can be important in tumor growth through interactions with stroma. The assessment of intra- and peritumoral stroma can be critical in the microscopic examination. Tumors with a high percentage of stroma (stroma-high) may have a large contact area between tumor cells and stroma. This point can be considered for the active interaction between tumor cells and stroma. Previous studies reported the prognostic impact of TSR in CRCs. However, the prognostic implications of TSR differ among studies in CRCs [2
]. Our results showed that stroma-low was significantly correlated with better OS and RFS in CRCs. However, Scheer et al. and Vogelaar et al. reported no significant correlation between TSR and survival in CRCs [2
]. Discrepant results can be caused by various factors, including different study populations, evaluation criteria, and methods.
We evaluated TSR using scoring percentages in 10% increments (10%, 20%, 30%, etc.). In the present study, additional analysis on the basis of 30% and 70% cutoffs was performed. In the subgroup analysis, according to evaluation criteria for stroma-low, a 30% cutoff but not a 70% cutoff had a prognostic implication in comparisons between TSR and OS and RFS. According to our results, lower cutoffs, such as 30% or 50%, are more effective than higher criteria in predicting the patient’s prognosis. In a previous study, Scheer et al. evaluated the prognostic impact of TSR by dividing their study sample into low (≤30%), intermediate (40%, 50%, and 60%), and high (≥70%) [13
]. There was no significant difference in survival between the intermediate-TSR and low-TSR (stroma-high) subgroups. In our study, patients were classified into four subgroups based on the following cutoffs: TSR ≤ 30%, 30–50%, 50–70%, and ≥70%. Patients with TSR ≤ 30% and 30–50% had similar survival rates. There was no significant difference in survival rates between patients with TSR 50–70% and ≥70%. According to these results, a 50% cutoff may be appropriate to distinguish stroma-low.
Histologic subtypes of CRCs include micropapillary, mucinous, and medullary carcinomas, etc. [22
]. The current study compared the TSR in CRCs with micropapillary and mucinous components. In our results, the rates of stroma-low differed among histologic subtypes. CRCs with a mucinous component showed a higher rate of stroma-low. However, CRCs with a micropapillary component were significantly correlated with stroma-high. This result suggested that TSR may be associated with the histologic characteristics of each subtype. The rule of assessment for TSR of the mucinous component has not been elucidated. In the present study, the mucinous component was evaluated as a tumor component in assessments of TSR. In a previous study, van Pelt GW et al. described that mucus, but not a mucinous component, should be ignored for scoring [23
]. However, Huijbers et al. reported that the mucinous component was included as the stromal area [24
]. If the entire tumor is composed of a mucinous component, the TSR is zero. This means that the tumor component is not present in tumors with TSR 0%. Moreover, mucinous components are included in mucin and floating tumor cells. The mucinous carcinoma is defined as a carcinoma with a greater than 50% mucinous component. Although the mucus without tumor cells is excluded, there was no specific reference for the interpretation of the mucinous component. Detailed information on the correlation between TSR and CRC with a mucinous component could not be obtained from previous studies [22
]. In addition, there was no explanation for the TSR assessment of the mucinous component and mucin pool when using an automatic deep learning algorithm [25
]. We previously reported the poor prognosis of CRCs with a micropapillary component [26
]. In our previous study, CRCs with a micropapillary feature were significantly correlated with vascular invasion. Further studies will be needed to determine the clinicopathological implication of intratumoral stroma in CRC with a micropapillary component.
According to our results, TSR was significantly correlated with vascular and perineural invasion and distant metastasis. In addition, stroma-low was significantly correlated with better survival in CRC. However, TSR was not associated with the pT stage and lymph node metastasis. Mesker et al. reported that TSR was useful in predicting prognosis for stage I–III colon cancers [10
]. Interestingly, in a patient with the pT4 stage, there was no significant correlation between TSR and OS and RFS (p
= 0.812 and p
= 0.915, respectively). However, there was a significant correlation between TSR and survival rate in the pT1–3 stages (OS, p
< 0.001; RFS, p
< 0.001). In addition, regardless of lymph node metastasis, stroma-low was significantly correlated with better OS and RFS. A detailed examination of vascular and perineural invasions in stroma-high tumors is needed.
As a result, Huijbers et al. reported that a correlation between stroma-low and vascular invasion was found [19
]. However, the relationship between the amount of stroma and vascular invasion is not fully understood. We investigated the MVD in the intratumoral stroma and the correlation between TSR and MVD in the present study. MVD was significantly higher in patients with stroma-high than in those with stroma-low. Also, the HIF-1α expression of tumor cells was significantly higher in the stroma-high subgroup than in the stroma-low subgroup. Additional subgroup analysis between high MVD and survival according to the TSR was added in the Supplementary Table (Table S1
). Pancreatic cancers with high stroma demonstrated high vessel density [27
]. Biologically, tumors with dense collagen content have good vascularization and are relatively well different [27
]. However, evidence could not be found for the correlation between HIF-1α expression and the amount of stroma in colorectal cancers. The correlation between intratumoral stroma and hypoxic status is not clear. However, CRCs with less intratumoral stroma have low-MVD and may be susceptible to hypoxia.