Assessment of compartment-specific CD103-positive cells for prognosis prediction of colorectal cancer

Abstract

Background

CD103⁺ tissue-resident memory T cells was detected in various solid malignancies, like colorectal cancer (CRC), and associated with improved survival. However, clinical significance of CD103⁺ cells in specific intratumor compartment remains unclear.

Methods

The abundance and distribution of CD103⁺ cells were assessed using immunohistochemistry and quantified separately for 3 compartments, including intraepithelial compartments at center of tumor (CT-IEL), stromal compartments at center of tumor (CT-ST) and invasive margin (IM) in a cohort of 224 CRC patients under radical surgery and correlated with outcome. Findings in each compartment were then validated in an external validation cohort comprising 294 CRC patients.

Results

Elevated density of CD103⁺ cells infiltration in the CT-IEL, CT-ST or IM compartment was correlated with favorable survival in both the initial discovery cohort and subsequent validation cohort. Notably, abundant CD103⁺ cells located in the CT-IEL compartment was remained an independent prognostic indicator for CRC patients by multivariant analysis. Characterization study showed that intraepithelial CD103⁺ cells were predominantly single positive CD8 T cells. Conversely, CD103⁺ cells exhibited a heterogeneous population comprising CD103⁺CD8⁺ cells, CD103⁺CD4⁺ cells, and nonconventional CD103⁺CD4⁺CD8⁺ cells in the CT-ST and IM compartments. Finally, a CD103 score was generated comprising abundance of CD103⁺ cells in the 3 compartments. This score had the highest relative contribution to the risk of all clinical parameters for prognosis in both cohorts.

Conclusion

This study supported a phenotypic heterogeneity of CD103⁺ cells in CRC, and provided a reliable estimate of the risk of death and recurrence in CRC patients based on combined analysis of CD103⁺ cells within 3 intratumor compartments.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-025-04087-z.

Introduction

Colorectal cancer (CRC) is a prevalent gastrointestinal malignancy, ranking as the second leading cause of tumor-related deaths globally [1–3]. Currently, the primary method utilized by healthcare professionals to assess the prognosis of CRC patients is the AJCC/TNM staging system and tumor cell differentiation [2]. Nevertheless, due to the significant heterogeneity, patients with identical tumor stages may result in disparate responses to identical treatments, ultimately yielding variable clinical outcomes. This highlights the limitations of anatomy-based staging systems in predicting prognosis [4]. There is increasing recognition in literatures that the infiltration of immune cells in CRC has been shown to be superior to standard staging systems as a prognostic factor [5, 6]. Furthermore, the remarkable efficacy of immunotherapy in anti-tumor treatment has reignited interest in the role of tumor-infiltrating immune cells in cancer research [7, 8].

Tissue-resident memory T cells (TRM) represent a unique type of long-lived memory T cells capable of residing in tissues and participating in rapid immune responses against invading pathogens [9]. The infiltration of TRM cells within tumors has been linked to enhanced survival rates in various solid malignancies [10]. Recent investigations have identified CD103 as a canonical marker for TRM cells [10]. CD8⁺ cells exposed to specific antigens within a TGF-β-enriched tumor microenvironment can upregulate CD103 expression and differentiate into CD8⁺ TRM cells. The interplay between CD103 and E-cadherin significantly influences the recruitment and stimulation of CD8⁺ TRM cells within tumor tissues, leading to the secretion of perforin and granzyme B and subsequently bolstering the anti-tumor immune response[11]. Studies have indicated that high infiltration of CD103⁺ cells correlate with improved clinical outcomes in ovarian cancer, breast cancer, lung cancer, and CRC [12–15]. It has been recognized that a significant of CD103⁺ cells in the tumor microenvironment concurrently express the CD8 molecule [10, 11, 16]. However, there are also different opinions that CD103 is also expressed at high levels on tumor-infiltrating regulatory T cells (Treg) and tumor-derived CD103⁺ Treg suppress CD8⁺ T cell responses more strongly than CD103⁻ Treg [17–19]. Furthermore, It is also known that CD103 is the marker for dendritic cells and tumor-infiltrating CD103⁺ dendritic cells are necessary for anti-tumor immunity [20]. Therefore, CD103⁺ cells exhibit significant heterogeneity within the tumor microenvironment.

Tumor-infiltrating immune cells exhibit significant spatial heterogeneity within a single tumor section. Research has indicated that CD8⁺ T cells located in the invasive margin (IM) are more numerous but also more functionally impaired compared to those found in the central of the tumor (CT) [21, 22]. Additionally, various immune cell populations such as CD8⁺, FOXP3⁺, and CD45RO⁺ T cells demonstrate varying densities and prognostic implications in the intraepithelial compartment (CT-IEL) and stromal compartment of the central tumor (CT-ST) [23–25]. Similarly, CD103⁺ cells in the CT-IEL compartment exhibited better survival than those in the CT-ST compartment of non-small cell lung cancer (NSCLC) [14]. Therefore, the anti-tumor efficacy of tumor-infiltrating immune cells is significantly influenced by their spatial localization and functional subpopulations. Previous study of CD103⁺ cells in CRC have utilized immunohistochemical tissue microarray (TMA) technique [15]. The tumor tissue areas of TMA cores were liminted, which hinder the precise quantification of intratumoral CD103⁺ cells in different compartments. Hence, a comprehensive characterization of CD103⁺ cell phenotypes and their functional roles within specific intratumoral compartments may provide insights into the immune functions of CD103⁺ cells in CRC.

Therefore, our study aimed to investigate the density, phenotype and prognostic value of CD103⁺ cells within specific CRC intratumor compartments. Additionally, we developed an CD103 score based on the infiltration of CD103⁺ cells in intratumor compartments to enhance the prognostic value of CD103⁺ cells in CRC.

Materials and methods

Patient cohorts

This study encompassed two independent cohorts of CRC patients. The discovery cohort consisted of 224 patients who received treatment for CRC between January 2009 and December 2012 in the Colorectal Surgery Department at the Sixth Affiliated Hospital of Sun Yat-sen University (SYSUSH). The validation cohort included 294 CRC patients treated at the Cancer Center of Sun Yat-sen University (SYSUCC) between January 2004 and September 2005. All patients in both cohorts underwent curative resection, exhibited no distant metastasis, and did not receive preoperative anti-cancer therapy. Cancer staging was conducted in accordance with the 7th edition of the TNM classification system established by the American Joint Committee on Cancer. Written informed consent was obtained from each patient, and the study was approved by the Clinical Research Ethics Committee of the Sixth Affiliated Hospital and the Cancer Center of Sun Yat-sen University. The primary objectives were to evaluate the prognostic significance of CD103⁺ cell density in relation to cancer-specific survival (CSS) and disease-free survival (DFS). CSS was defined as the duration from tumor resection to death attributable to cancer. DFS was defined as the period from the date of surgery to first observation of disease recurrent.

Immunohistochemistry

The CRC samples were sectioned into 4 μm slices and used for immunohistochemical staining. The tissue sections underwent deparaffinization in xylene followed by rehydration through a graded series of alcohol solutions. Antigen retrieval was performed via microwave treatment in Tris–HCl buffer (pH 9.2). To inhibit endogenous peroxidase activity, the sections were treated with a 0.3% hydrogen peroxide (H₂O₂) solution. The sections were then blocked with goat serum at room temperature for 30 min and incubated with a monoclonal anti-CD103 primary antibody (Abcam) overnight at 4 °C. Subsequently, the sections were incubated with horseradish peroxidase-conjugated secondary antibodies (DAKO) at 37 °C for 30 min. Detection was performed using diaminobenzidine, followed by counterstaining with hematoxylin. Finally, the sections underwent washing, dehydration, and fixation.

Immunohistology evaluation

The density of CD103⁺ cells within CT-IEL, CT-ST, and IM was analyzed by two independent observers who were blinded to patient outcomes and other clinical findings. The areas of CRC tissue were initially screened at low power (100 × magnification), and the five most representative high-power fields were subsequently captured at 400 × magnification for distinct tumor compartments in each specimen. The total number of CD103⁺ infiltrating cells in each field was counted manually, and the mean value across five fields was calculated to determine the score for each tumor compartment. Finally, the average of the scores from the two independent observers was computed and employed as the definitive value for statistical analysis. To mitigate the impact of inter-cohort variation, we determined independent, cohort-specific thresholds based on the density of CD103⁺ cells for each cohort. The cutoff value was defined as the lower tertile of CD103⁺ cell density. In the discovery cohort, the cutoff values for CD103⁺ cells in the CT-IEL, CT-ST, and IM compartments were 17.40, 26.90, and 38.30 per field, respectively. In the validation cohort, the corresponding cutoff values were 14.10, 20.00, and 26.80 per field, respectively.

Multiplex immunofluorescence staining

For multiplex immunofluorescence (mIF), the preparation and antigen retrieval of slides were conducted following standard immunohistochemistry protocols. The tyramide signal amplification (TSA) 4-Color IHC kit (abs50012-20T; Absin) was employed for mIF staining. Each slide underwent multiple rounds of staining, beginning with an overnight incubation of the primary antibody at 4℃. This was followed by a 30-min incubation with the corresponding secondary antibodies at 37℃, and subsequent covalent binding of TSA fluorophores. Specifically, anti-CD103 (Abcam), anti-CD4 (Abcam), and anti-CD8 (CST) antibodies were conjugated with TSA dye 520, TSA dye 570, and TSA dye 650, respectively. Nuclei were counterstained with DAPI.

Multiplex immunofluorescence imaging acquisition and analysis

The mIF slides were imaged and analyzed utilizing the TissueGnostics system. The composite multispectral images were decomposed into individual spectral images. Subsequently, appropriate positive thresholds for each biomarker were determined according to signal intensity, followed by the elimination of non-specific fluorescence. The correct nuclear signals were screened and then cell colocalization was performed to identify positive cells based on the membrane signal intensity of each biomarker. On this basis, the three most representative fields of each microcompartments were analyzed and counted all immune cell infiltrations per field.

CD103 score

Accorrding to the density of CD103⁺ cells within each intratumoral compartment, patients were stratified into Int + Hi and Lo groups in both the discovery and validation cohorts. Subsequently, data from the CT-IEL, CT-ST, and IM compartments were consolidated and converted into a CD103 scoring system. Patients exhibiting Int + Hi density in all three compartments were classified as CD103 score of Hi. Those with Int + Hi density in any one or two of the compartments were classified as CD103 score of Int. Patients with low CD103⁺ cell infiltration across all three compartments were classified as CD103 score of Lo.

Statistical analysis

The Kaplan–Meier method and log-rank test were utilized to evaluate survival outcomes between two distinct patient subgroups. Additionally, Cox proportional hazards models were employed to examine the impact of various covariates on survival, including CD103⁺ cell density in different compartments, age, gender, histological type, histological grade, tumor location, T status, N status, vessel invasion, and perineural invasion. Hazard ratios (HR) and 95% confidence intervals (CI) for multivariate analyses were calculated using Cox proportional hazards regression models. Spearman correlation tests were used to analyze the associations of clinicopathological variables between the discovery and validation cohorts. The relationships between CD103⁺ cell density and clinicopathological variables were assessed using the t-test. Multiple correlations of CD103⁺ cell density across three intratumoral compartments were analyzed utilizing the multiple R-squared test. Chi-squared proportion tests were conducted to evaluate the relative contribution of each parameter to survival risk in CSS and DFS. All statistical tests were two-sided, with a significance threshold set at P < 0.05. The statistical analyses were performed using GraphPad Prism (version 8.02) and IBM SPSS Statistics (version 22) software.

Results

Patient characteristics

This study enrolled a total of 663 patients diagnosed with AJCC/UICC TNM stage I-III CRC, comprising 307 patients from SYSUSH and 356 patients from SYSUCC. A total of 145 patients were excluded from the analysis due to the absence of tumor paraffin wax in 65 patients and the lack of representative CT-ST or IM compartments in 80 patients. Consequently, 224 patients from SYSUSH were enrolled into the discovery cohort, while 294 patients from SYSUCC were assigned to the validation cohort (Fig. 1a). The demographic and clinical characteristics of the patients were well balanced between the discovery and validation cohorts, with the exception of the number of cases of mucinous adenocarcinoma (Table S1). Overall, patients were 55.8% male, and the median age was 60 years ([IQR] 51–69). Among the whole cohort, 101 patients (19.5%) were diagnosed with stage I CRC, 217 patients (41.9%) with stage II CRC, and 200 patients (38.6%) with stage III CRC. Across all patients, 119 (23.0%) experienced relapses, and 98 (18.9%) succumbed to CRC-related causes. The median follow-up time was 101.8 months for the discovery cohort and 106.5 months for the validation cohort. The median survival time from surgery to CRC-related death was 38.6 months.

Fig. 1.

Study design and representative images of CD103⁺ cell infiltration in CRC tissues. a Study flowchart. b The upper part of the image illustrated the representative divisions of the CT-IEL, CT-ST, and IM compartments within CRC tissues. The lower part of the images presented representative examples of each intratumoral compartment utilized for immunohistological evaluation

The density of CD103⁺cells in different intratumoral compartments

To elucidate the density of CD103⁺ cells, the tumor specimen of each patient was classified into three distinct regions: the intraepithelial compartments at the center of the tumor (CT-IEL), the stromal compartments at the center of the tumor (CT-ST), and the invasive margin (IM) (Fig. 1b). In the discovery cohort, the density of CD103⁺ cells was found to be highest in the IM compartment, whereas the CT-IEL compartment exhibited the lowest density of CD103⁺ cells among the three intratumoral compartments (Fig. 2a). Similar results were observed in both the validation cohort and the whole cohort; however, no significant difference was found in the density of CD103⁺ cells between CT-IEL and CT-ST within the validation cohort (Fig. 2b, c). Additionally, the density of CD103⁺ cells across the three intratumoral compartments demonstrated a strong correlation within the various patient cohorts (Fig. 2a, b, c).

Fig. 2.

The densities of CD103⁺ cells in different compartments and prognostic value of clinicopathologic characteristics in patient cohorts. a, b, c The violin plots illustrated the densities of CD103⁺ cells across distinct compartments for each patient cohort. a, b, c The scatter plots depicted the multivariate correlation of CD103⁺ cells among three intratumoral compartments across various patient cohorts. d, e The forest plots represented the prognostic significance of clinicopathologic characteristics on CSS and DFS within each patient cohort. *p < 0.05, **p < 0.01, ***p < 0.001

Associations between the density of CD103⁺ cells and clinicopathologic characteristics are listed in Table S2and Table S3. In the discovery cohort, tumors with a higher N stage or positive perineural invasion exhibited lower density of CD103⁺ cells across all three intratumoral compartments. The density of CD103⁺ cells in the CT-IEL and IM compartments were significantly reduced in patients who experienced postoperative recurrence and metastasis, a pattern not observed in the CT-ST compartment. In the CT-ST compartment, patients with a lower T stage demonstrated higher infiltration of CD103⁺ cells. Mucinous adenocarcinoma exhibited lower infiltration of CD103⁺ cells in the IM compartment. Unexpectedly, patient gender was correlated with the density of CD103⁺ cells in CT-IEL compartment. In the validation cohort, patients presenting with higher T stage, higher N stage, or postoperative recurrence and metastasis demonstrated lower density of CD103⁺ cells across all three intratumoral compartments. Mucinous adenocarcinoma showed diminished infiltration of CD103⁺ cells in both the CT-ST and IM compartments. Additionally, poorly differentiated tumors were associated with lower infiltration of CD103⁺ cells in the CT-ST compartment. No significant associations were observed between the density of CD103⁺ cells with age or gender of patients in validation cohort.

Prognostic value of clinicopathologic characteristics in patient cohorts

Among the clinicopathologic characteristics of the discovery cohort, a higher N stage and positive vascular invasion were strongly associated with worse CSS (Fig. 2d). In terms of DFS, female CRC patients exhibited a survival advantage compared to males, while a higher N stage remained a significant risk factor (Fig. 2e). In the validation cohort, patients with higher T and N stages presented significantly worse outcomes for CSS (Fig. 2d). In terms of DFS, patients with right-sided colon cancer had better outcomes and a higher N stage predict a worse risk for DFS (Fig. 2e). Across the whole cohort, tumor location and N stage were strongly associated with survival outcomes (Fig. 2d, e).

Prognostic value of CD103⁺cells within different intratumoral compartments on the survival of patients with CRC.

Based on the density of CD103⁺ cells in each compartment, patients were stratified into three groups: high infiltration (Hi), intermediate infiltration (Int), and low infiltration (Lo), from highest to lowest (Table S4). The three-category Kaplan–Meier analysis identified patients with distinct clinical outcomes for CSS and DFS. However, no significant differences were observed between the Hi and Int groups across most subgroups (data not shown). Consequently, the Int and Hi groups were combined into a new group, designated as the Int + Hi group. The lower tertile of CD103⁺ cell density was used as the cutoff value for each compartment. In the CT-IEL compartment, patients with high and intermediate levels of CD103⁺ cell infiltration presented significantly better outcomes for CSS (Fig. 3a, p = 0.0005) and DFS (Fig. 3d, p = 0.0016) compared to patients with low levels of CD103⁺ cell infiltration. Similarly, in the CT-ST compartment, patients in the Int + Hi group had a survival advantage in both CSS (Fig. 3b, p = 0.0138) and DFS (Fig. 3e, p = 0.0224) compared to those in the Lo group. In the IM compartment, high and intermediate CD103⁺ cell infiltration was positively associated with CSS (Fig. 3c, p = 0.0131), while no significant effect was observed on DFS (Fig. 3f, p = 0.1444). Within the discovery cohort, a multivariate model was utilized to identify independent prognostic factors. This model encompassed all clinicopathological variables, along with various compartments of CD103⁺ cell density, which exhibited significant prognostic value in the univariate analysis. The results indicated that CD103⁺ cell infiltration in the CT-IEL compartment and vascular invasion independently held prognostic significance for CSS (Fig. 3g). Subsequently, we analysed the relative contribution of each parameter to predict CSS and found that the most important variables were vascular invasion (46.0%) and CD103⁺ cells in the CT-IEL compartment (36.9%, Fig. 3h). However, no variable exhibited independent prognostic significance for DFS (Fig. 3g).

Fig. 3.

The impact of different compartments of CD103⁺ cell density on the outcomes of CRC patients in discovery cohort. Kaplan–Meier curves illustrated a beneficial effect of CD103⁺ cell density in the CT-IEL (a, d) and CT-ST (b, e) compartments on CSS and DFS. In the IM compartment, CD103⁺ cell density exhibited a positive effect on CSS, but not on DFS (c, f). g A Cox multivariable regression analysis of CSS and DFS was conducted, incorporating CD103⁺ cells in the CT-IEL, CT-ST, and IM compartments alongside clinical parameters such as gender, N stage, and vascular invasion. h The relative contribution of each parameter to survival risk for CSS and DFS was assessed using a chi-squared proportion test. *p < 0.05, **p < 0.01, ***p < 0.001

To validate the findings derived from the discovery cohort, we incorporated a validation cohort comprising 294 CRC patients. Analogous to the discovery cohort, patients in the validation cohort were categorized into Int + Hi and Lo groups based on the density of CD103⁺ cells. Patients exhibiting high and intermediate infiltration of CD103⁺ cells within the CT-IEL compartment demonstrated significantly prolonged survival for both CSS (Fig. 4a, p < 0.0001) and DFS (Fig. 4d, p < 0.0001). In the CT-ST compartment, the Int + Hi group significantly predicted a lower risk for CSS (Fig. 4b, p = 0.0001) and DFS (Fig. 4e, p < 0.0001) compared to the Lo group. Comparable outcomes were observed in the IM compartment, where the Int + Hi group demonstrated a positive impact on CSS (Fig. 4c, p = 0.0001) and DFS (Fig. 4f, p < 0.0001). In the multivariate analysis, the infiltration of CD103⁺ cells in the CT-IEL compartment remained significant for CSS and DFS within a multivariable Cox model that included tumor location, T stage, N stage, and CD103⁺ cell infiltration in three compartments (Fig. 4g). The variables with the most important relative contribution to the risk were as follows: for CSS, N stage (55.5%) and CD103⁺ cells in CT-IEL (37.7%); for DFS, N stage (53.2%), CD103⁺ cells in CT-IEL (25.1%), and location (15.3%) (Fig. 4h). Furthermore, we also analyzed the prognostic value of CD103⁺ cells in the whole cohort and found that the most important variables were CD103⁺ cells in CT-IEL (53.3%), N stage (25.1%), and location (16.1%) in CSS, and CD103⁺ cells in CT-IEL (52.3%), N stage (25.7%), and location in DFS (17.2%, Figure S2).

Fig. 4.

Prognostic value of CD103⁺ cell density in a validation cohort of CRC patients. The Kaplan–Meier survival analysis showed the probability of CSS and DFS, stratified according to the density of CD103⁺ cells within the compartments of CT-IEL (a, d), CT-ST (b, e), and IM (c, f). g Cox multivariable regression analysis of CSS and DFS included CD103⁺ cells in the CT-IEL, CT-ST, and IM compartments, along with clinical parameters such as T stage, N stage and location. h The relative contribution of each parameter to survival risk for CSS and DFS was assessed using a chi-squared proportion test. *p < 0.05, **p < 0.01, ***p < 0.001

Immunolocalization of different CD103⁺cell subpopulations in the intratumoral compartments of CRC tissues

CD103⁺ cells, similar to many other tumor-infiltrating immune cells, exhibit significant heterogeneity within tumor tissues, with only certain subsets demonstrating anti-tumor activity. One study has indicated that the infiltration of CD103⁺CD4⁺ T cells is associated with lower survival rates in gastric cancer [26]. To identify the subset of CD103⁺ cells across intratumoral compartments, we performed colocalization studies and calculated the frequency of CD103⁺ cells co-expressing the T helper cells marker CD4 and the cytotoxic T cells marker CD8. The mIF staining revealed a substantial infiltration of both CD103⁺ cells and CD8⁺ T cells in the CT-IEL compartment, with approximately 72.2% of CD103⁺ cells being CD8⁺CD103⁺ double-positive (DP) cells. In contrast, CD4⁺ T cells in the CT-IEL compartment were scarce, with only 10.2% of CD103⁺ cells co-stained with CD4 (Fig. 5a). Within the CT-ST compartment, approximately half (57.4%) of the CD103⁺ cells were CD8⁺CD103⁺ DP cells. The frequencies of CD4⁺CD103⁺ DP cells were 14.9%. In addition, we identified a small population of CD4⁺CD8⁺CD103⁺ triple-positive (TP) cells, constituting less than 5% of the total CD103⁺ cell population (Fig. 5b). In the IM compartment, the proportion of colocalized cells among the total CD103⁺ cells was comparable to that observed in the CT-ST compartment (Fig. 5c). Notably, the infiltration of CD8⁺CD103⁺ DP cells was higher in the CT-IEL compartment compared to the CT-ST and IM compartments (Fig. 5d). Conversely, the proportions of CD4⁺CD103⁺ DP cells and TP cells were lower in the CT-IEL compartment compared to the CT-ST and IM compartments (Fig. 4e, f). Collectively, these findings indicate that the CT-IEL compartment is predominantly characterized by a CD8⁺CD103⁺ DP lymphocyte phenotype among CD103⁺ cells, whereas the CT-ST and IM compartments exhibit a more complex subpopulations of CD103⁺ cells.

Fig. 5.

Immunolocalization of different CD103⁺ cell subpopulations in the intratumoral compartments of CRC tissues. The representative mIF staining illustrated the infiltration of CD103 (green), the T helper cell marker CD4 (violet), and the cytotoxic T cell marker CD8 (red) within the CT-IEL (a), CT-ST (b), and IM (c) compartments of CRC tissues. The proportion of cells positive for these markers relative to the total number of CD103⁺ cells is depicted for the CT-IEL (a), CT-ST (b), and IM (c) compartments of CRC tissues. (d) The proportion of CD8⁺CD103⁺ DP cells was observed to be higher in the CT-IEL compartment compared to the CT-ST and IM compartments. The proportions of CD4⁺CD103⁺ DP cells (e) and CD4⁺CD8⁺CD103⁺ TP cells (f) were found to be lower in the CT-IEL compartment relative to the CT-ST and IM compartments. Scale bars indicate 50μm. *p < 0.05, **p < 0.01, ***p < 0.001

Combined assessment of CD103⁺cells density in different intratumoral compartments to predict clinical outcome

To differentiate patient subgroups with distinct clinical outcomes, CRC patients were categorized based on a CD103 score for CD103⁺ cells within each intratumoral compartment (CT-IEL, CT-ST, IM). Patients were stratified into Int + Hi and Lo groups within each compartment according to the classifications established in the original cohort. Subsequently, data from CT-IEL, CT-ST, and IM were consolidated and transformed into a composite CD103 score. Using this composite CD103 score, three distinct CRC patient subgroups were identified (Lo, Int, and Hi). Patients exhibiting low infiltration of CD103⁺ cells across all three tumor compartments were designated as having a CD103 score of Lo. Conversely, patients with intermediate to high density in all three compartments were assigned a CD103 score of Hi. Those with Int + Hi density in one or two of the three compartments were categorized with a CD103 score of Int. Based on this classification, 18.15, 31.47, and 50.39% of the whole cohort of CRC patients were classified into the Lo, Int, and Hi groups, respectively (Fig. 6a). Notably, patients with a CD103 score of Hi demonstrated the highest 5-year CSS rate at 92.0% and DFS rate at 88.5%. The Lo and Int groups presented with CSS and DFS rates at 5 years of 83.1 and 73.7, and 71.4 and 60.7%, respectively. Patients with a low CD103 score had shorter survival times. Kaplan–Meier analysis identified that patients in Lo group illustrated the significantly shorter CSS (Fig. 6b) and DFS (Fig. 6c) compared to those in the Int and Hi groups. Furthermore, patients with a high CD103 score exhibited significantly longer survival for both CSS and DFS compared to patients in the Int group. Utilizing Cox multivariate regression analysis, it was determined that the CD103 score, N stage, and tumor location were significantly and independently associated with CSS and DFS (Fig. 6d). The power of CD103 score to predict CSS and DFS was superior to that of the N stage and tumor location (Fig. 6e). These findings were consistent across both the discovery and validation cohorts, establishing the CD103 score as an independent prognostic factor for patients with CRC. Furthermore, the relative contribution to the risk showed that CD103 score was better than all clinical variables (Figure S3, S4).

Fig. 6.

The impact of CD103 score on the outcome of whole cohort of CRC patients. Kaplan–Meier survival analysis was conducted to assess CSS and DFS in 518 CRC patients, stratified by the densities of CD103⁺ cells across various intratumoral compartments (CT-IEL, CT-ST, and IM). a The pie charts illustrated the proportion of the three CD103 score groups within whole cohort. b, c The Kaplan–Meier survival analysis showed the probability of CSS and DFS stratified by CD103 score. d Cox multivariable regression analysis of CSS and DFS included CD103 score, N stage and location. e The relative contribution of each parameter to survival risk for CSS and DFS was assessed using a chi-squared proportion test for CD103 score, N stage, and location. The pie charts illustrated the proportions of the three CD103 score groups among patients without (f) and with lymph node metastasis (i). The Kaplan–Meier curves represented the probability of CSS and DFS in relation to the CD103 score among patients without (g, h) and with lymph node metastasis (j, k). *p < 0.05, **p < 0.01, ***p < 0.001

Prognostic value of CD103 score among patients with or without lymph node metastasis

In our research, lymph node metastasis is identified as an independent high-risk factor for patients in the validation cohort and the whole cohort. To assess the prognostic impact of CD103 score in patients without lymph node metastases (stage I-II CRC) and in those with lymph node metastases (stage III CRC), subgroup analyses were conducted based on N stage. The proportion of patients in the N stage subgroups was found to be similar to that of the overall group (Fig. 6f, i; overall v N = 0, p = 0.7362; overall v N = 1,2, p = 0.5906). In both N stage subgroups, a high CD103 score was correlated with longer survival among CRC patients. Lymph node negative patients with Hi CD103 score exhibited markedly superior survival outcomes for both CSS and DFS in comparison to those with Int and Lo CD103 scores (Fig. 6g, h). The intermediate group also displayed a greater impact on CSS and DFS than the Lo group. In patients with positive lymph node status, the high CD103 score group significantly associated with a lower risk for CSS and DFS compared to the Int and Lo CD103 score groups (Fig. 6j, k). However, there was no significant difference in CSS and DFS survival curves between the Int and Lo CD103 score groups. Therefore, CD103 score emerged as a significant predictor of survival in both stage I-II and stage III CRC patients.

Discussion

Our study included two independent cohorts from two hospitals, comprising a collective 518 patients with Stage I-III CRC. Elevated levels of CD103⁺ cell infiltration in the CT-IEL, CT-ST, and IM compartments were found to be associated with improved survival outcomes in discovery and validation cohorts. CD103⁺ cells located in CT-IEL compartment could serve as an independent prognostic factor for patients with CRC. Furthermore, the majority of CD103⁺ cells within the CT-IEL compartment exhibited a CD8⁺CD103⁺ cell phenotype, whereas in the CT-ST and IM compartments, CD103⁺ cells represented a heterogeneous population of cell phenotypes. Additionally, an CD103 score was formulated based on the infiltration of CD103⁺ cells in intratumoral compartments, facilitating the delineation of patient subgroups with diverse clinical outcomes. These results imply that CD103⁺ cells in various intratumor compartments, particularly the CT-IEL compartment, may serve as prognostic indicators for patients with CRC.

Previous research has demonstrated a significant correlation between CD103⁺ cells and survival in NSCLC and ovarian cancer, without the need to discriminate their location in intraepithelial versus stromal compartments [12, 14]. Consistent with these findings, our study revealed that the infiltration of CD103⁺ cells across all three intratumor compartments is associated with enhanced survival outcomes in two independent cohorts of CRC patients. The power of CD103⁺ cells in the CT-IEL compartment for predicting survival was superior to that of CD103⁺ cells in the CT-ST and IM compartments. Notably, only CD103⁺ cells in the CT-IEL compartment function as an independent prognostic factor for patients with CRC. Similar with our findings, CD103⁺ cells within the intraepithelial compartment also serve as an independent prognostic factor in NSCLC and urothelial cell carcinoma of the bladder (UCB) [14, 27]. The protective role of CD103⁺ cells in the intraepithelial compartment of different tumors, along with their superior clinical survivals, may be attributed to their epithelium-favored infiltration pattern. E-cadherin, the natural ligand for CD103, facilitates the persistent residency and anti-tumor immune response of CD103⁺CD8⁺ cells [11]. Our study found a positive correlation between CD103⁺ cells and E-cadherin expression within the intraepithelial compartment (Figure S5), which is consistent with previous studies in UCB and prostate cancer [27, 28].

The intratumoral distribution and prognostic value of CD103⁺ cells are also influenced by their functional subpopulations. Previous studies have showed coexpression of CD103 and CD8 in CRC tissues [15]. In our study, we observed that the majority of CD103⁺ cells within the CT-IEL compartment were CD103⁺CD8⁺ cells. Conversely, the CT-ST and IM compartments exhibited a heterogeneous population comprising CD103⁺CD8⁺ cells, CD103⁺CD4⁺ cells, and nonconventional CD103⁺CD4⁺CD8⁺ cells. It is widely accepted that mature peripheral T cells express either CD4 or CD8, but not both. Within the tumor microenvironment, CD8⁺ T cells predominantly engage in anti-tumor cytotoxic activities, while CD4⁺ T cells perform helper functions to regulate the immune response. However, mature CD4⁺CD8⁺ T cells have been identified in various tumor tissues, and their origin and biological functions remain largely unknow [29]. A study on melanoma revealed that intratumoral CD4⁺CD8⁺ T cells originate from CD8⁺ T cells and exhibit transitional characteristics from a cytotoxic effect to a helper function [30]. In renal cell carcinoma, CD4⁺CD8⁺ T cells represent a subpopulation of CD8⁺ T cells characterized by the upregulation of CD4 expression. This observation led the authors to propose that these cells are the result of an antigen-driven expansion of CD8⁺ T cells and subsequently differentiation into a CD4⁺CD8⁺ T cell phenotype [31]. Furthermore, it has been reported that metastatic CRC samples exhibit a higher proportion of CD4⁺CD8⁺ T cells compared to non-metastatic CRC samples and normal colonic mucosa. These cells have been reported to promote tumor growth or metastasis and/or suppress immune responses in CRC through their high capacity to secrete IL-4 and IL-13 [32]. Another study identified that there are a population of T cells co-expresses both CD4 and CD8a in the human colonic mucosa and blood, which displays a type 1-like T regulatory cells phenotype and play a role in colon homeostasis [33]. In connection with the aforementioned literature, we hypothesize that these CD103⁺CD4⁺CD8⁺ cells may be more closely associated with immunosuppressive or immune regulatory functions rather than direct anti-tumor immune effects. The expression of CD103 indicates a tumor-resident phenotype of CD4⁺CD8⁺ T cells, suggesting that these cells are mature peripheral T cells rather than naïve CD4⁺CD8⁺ T cells. In addition, one study has shown that the infiltration of intratumoral CD103⁺CD4⁺ T cells is associated with decreased survival rates in gastric cancer [26]. Therefore, CD103⁺ cells in the CT-ST and IM compartments showed less prognostic contribution compared to CD103⁺ cells in the CT-IEL compartment, potentially due to the heterogeneous immune cell infiltration within the CT-ST and IM compartments.

There is a growing body of evidence indicating that the immune cells have a profound impact on the prognosis of tumor patients. Galon and colleagues have developed a standardized method for assessing lymphocyte immune infiltration, which entails analyzing the combined populations of CD8⁺ and CD3⁺ cells within both the tumor core and the invasive margin [34, 35]. This scoring system effectively categorizes patients into high- and low-risk groups, demonstrating significant discrepancies in time to recurrence, disease-free survival, and overall survival. Galon selected the T cell marker CD3 and cytotoxic T cell marker CD8 for their staining quality and antigen stability. With a deeper understanding of the tumor immune microenvironment, a potential future immune scoring system could encompass multiple immune parameters, including markers for cytotoxic T cells, memory T cells, T cell exhaustion, and immune suppression. To enhance the prognostic value of CD103⁺ cells in CRC, we devised an CD103 score based on the presence of CD103⁺ cells in distinct intratumor compartments. In the discovery cohort and validation cohort, the relative contribution of a single compartment of CD103⁺ cells alone to predict survival was found to be inferior to that of vascular invasion or N stage. Nonetheless, a combined analysis of CD103⁺ cells in intratumor compartments showed superior predictive power than that of vascular invasion and N stage. Specifically, the power of the CD103 score to forecast CSS and DFS was nearly five times greater than that of the N stage in the whole cohort. In high-risk patients with lymph node metastasis, CD103 score also allowed the identification of patients with distinct clinical outcome for CSS and DFS. Patients with an Hi CD103 score exhibited significantly improved survival compared to those in the other two groups. However, the survival curves of patients in the Int and Lo CD103 score groups overlapped, indicating that the presence of only one or two compartments with high CD103⁺ cell infiltration may not be sufficient to accurately predict survival in this patient population. Thus, an integrated examination of CD103⁺ cells across different intratumor compartments revealed improved predictive value for survival outcomes among diverse CRC patient groups compared to analyzing a single compartment alone. However, the potential limitation of this CD103 score may arise from its dependence on a single immune parameter. In agreement with our findings, studies have showed that a significant of CD103⁺ cells in the tumor microenvironment concurrently express the CD8 molecule [16]. Furthermore, CD103⁺CD8⁺ cells demonstrated a more pronounced tissue-resident signature and anti-tumor efficacy in comparison to CD103⁻CD8⁺ cells [36]. The infiltration of intratumor CD103⁺ cells presented a superior prognostic value than CD8⁺ cells in various cancers [12, 14, 27, 37]. Hence, a combined evaluation of CD103, rather than CD8, along with other immune parameters in the CT-IEL, CT-ST, and IM compartments may serve as a valuable immune scoring system for assessing the degree of antitumor immunity.

Our study has certain limitations, including the reliance on the pathologist's clinical experience and personal discretion in selecting representative compartments for CD103⁺ cell infiltration. This approach may introduce selectivity bias and inefficiency, posing challenges for clinical applicability. Incorporating machine learning techniques to identify and quantify CD103⁺ cell infiltration could potentially mitigate these limitations and improve the accuracy and efficiency of the analysis. Moreover, a limited number of patients underwent testing for MSS/MSI status, which hindered a comprehensive evaluation of the impact of MSS/MSI status on the infiltration of CD103⁺ cells. Additionally, as these patients did not receive any form of immunotherapy, the potential role of CD103⁺ cells in intratumor compartments in predicting response to immunotherapy remained uncertain.

To conclude, our present study investigated the intratumor compartment and phenotype of CD103⁺ cells within CRC tissues, suggesting that the infiltration of CD103⁺ cells in various intratumor compartments could be a significant prognostic factor for CRC patients. Tumor-specific CD103⁺CD8⁺ TRM cells are crucial in the anti-tumor immune response, but further investigation is needed to elucidate their precise role in tumor biology, which appears to be influenced by their intratumor locations and activation status. Understanding this mechanism could provide valuable insights for the development of immunotherapy strategies for colorectal cancer.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CRC

Colorectal cancer

CSS

Cancer-specific survival

CT-IEL

The intraepithelial compartments at center of tumor

CT-ST

The stromal compartments at center of tumor

DFS

Disease-free survival

IM

Invasive margin

mIF

Multiplex immunofluorescence

TRM

Tissue-resident memory T cells

Author’s contribution

Guarantor of integrity of the entire study: MH and PH. Manuscript preparation: AH, YW, MA and HY. Experiments and data analysis: AH, YW and YX. Sample collection: JC, MC and JZ. Manuscript revision: MH and PH. All authors read and approved the final manuscript.

Funding

Support for these studies was provided by the Natural Science Foundation of Guangdong Province (2022A1515012457, JC), the Sixth Affiliated Hospital of Sun Yat-sen University Clinical Research-'1010' Program (1010CG(2022)-02, MH), the Fundamental Research Funds for the Central Universities, Sun Yat-sen University (2022007, MH).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethics approval and consent to participate

Written informed consent was obtained from each patient, and the study was approved by the Clinical Research Ethics Committee of the Sixth Affiliated Hospital and the Cancer Center of Sun Yat-sen University.

Consent for publication

All the authors agree to publish this paper.