Clinicopathologic, genetic and immune cell infiltration analysis of colorectal signet ring cell carcinoma with comparison to conventional adenocarcinoma

Abstract

Objective

We sought to characterize the genomic and immune landscape of signet ring cell carcinoma (SRCC), a rare and aggressive subtype of colorectal cancer (CRC).

Methods

Tissue samples and clinicopathological data were retrospectively analyzed from 37 SRCC and 172 conventional adenocarcinomas (AC) of rectum and sigmoid colon. The genetic and immune profiles were assessed using DNA next-generation sequencing (NGS) and multiplex immunohistochemistry.

Results

Compared to AC, SRCC patients were younger, had higher tumor, node, metastasis (TNM) stages and tumor grades (all P < 0.05), and exhibited significantly worse 3-year disease-free survival (DFS) and overall survival (OS) (both P < 0.0001). SRCC exhibited fewer somatic mutations in well-known CRC driver genes including APC (32 % vs. 74 %), KRAS (16 % vs. 42 %), and FBXW7 (0 vs. 20 %), but showed higher frequencies of genetic alterations in SMAD4, RNF43, BCL2L11 (present only in SRCC), MYC, and ARID2 (all P < 0.05). SRCC showed significantly higher levels of CD8⁺ tumor-infiltrating lymphocytes (TILs), but lower PD-1⁺CD8⁺ TILs in both stroma and intratumoral regions (all P < 0.05). Interestingly, high PD-1⁺CD8⁺ TILs in intratumoral regions significantly predicted longer DFS and OS in all SRCC and stage II-III SRCC patients (all P < 0.05), while CD3⁺ or CD8⁺ TILs did not. Moreover, the characteristic immune cell infiltration correlated with specific genetic alterations in SRCC.

Conclusions

Colorectal SRCC is a clinically and molecularly distinct and highly malignant subtype compared to AC. It exhibits a “pseudo-T cell-inflamed” tumor microenvironment with increased total CD8⁺ TILs but reduced PD-1⁺CD8⁺ TIL infiltration. Notably, PD-1⁺CD8⁺ TIL infiltration is associated with favorable prognosis. These findings provide new insights into tumor-immune interactions in SRCC, with potential implications for prognostic stratification and the development of personalized immunotherapeutic strategies.

1. Introduction

Colorectal cancer (CRC) is a heterogeneous disease with diverse tumorigenesis, immunogenicity, and clinical outcomes. Although the overall incidence of CRC has declined in recent decades, the incidence of early-onset CRC (diagnosed before age 50) is increasing, with signet ring cell carcinoma (SRCC) being relatively more prevalent in this group.^1,2 Compared to conventional adenocarcinoma (AC), SRCC is highly aggressive and has a poorer prognosis.^2,3 Despite notable advances in targeted therapies and immunotherapies, their efficacy in SRCC remains unsatisfactory, highlighting the urgent need for novel treatment strategies tailored to this subtype.

The tumor microenvironment (TME) plays a pivotal role in the prognosis of CRC, with immune infiltration patterns serving as important prognostic indicators. The Immunoscore (IS), based on CD3⁺ and CD8⁺ tumor-infiltrating lymphocytes (TILs), has been shown to outperform the traditional TNM staging system or microsatellite instability (MSI) status in predicting outcomes in CRC.^4,5 Immune checkpoints (ICs), such as programmed cell death protein 1 (PD-1), have also emerged as prognostic biomarkers in several malignancies. Although conflicting results exist, recent studies suggest that the infiltration of PD-1 expressing CD3⁺ and CD8⁺ TILs may be associated with favorable prognosis.6, 7, 8, 9, 10

However, the immunogenomic characteristics of colorectal SRCC remain poorly characterized. In particular, few studies have examined its immune landscape in comparison with AC, and the prognostic significance of TILs in SRCC remains largely unexplored. A better understanding of these features may help identify distinct immune signatures or treatment targets, paving the way for specific therapeutic strategies for colorectal SRCC.

In this study, we comprehensively characterized the clinicopathologic and genetic features of colorectal SRCC and compared them with those of AC. Using multiplex immunohistochemistry, we examined immune cell infiltration patterns and their associations with clinicopathologic and genomic features. This study aims to elucidate the tumor-specific biological features underlying the aggressive behavior of SRCC, explore potential pathogenic mechanisms, and provide insights for more effective and personalized therapeutic strategies for this challenging subtype.

2. Materials and methods

2.1. Study design, data and sample collection

Clinical data and tissue samples from 37 patients with rectal and sigmoid colon SRCC treated at Peking Union Medical College Hospital (PUMCH) between 2012 and 2023 were retrieved from the institutional electronic database and the Department of Pathology’s archival repository. As a control group, 172 cases of rectal and sigmoid colon AC that were managed at the Department of General Surgery during the same time period and had undergone next-generation sequencing (NGS) using either the 520-gene or 1021-gene panel were selected. All tissue samples were paraffin-embedded pre-treatment specimens, with tumor cell content ≥10 %. Cases of multiple primary CRC were excluded. All tissue samples were independently reviewed by two pathologists. Clinical and pathological data were collected, including age, sex, serum levels of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9), tumor size, histological grade, TNM stage, metastasis, and risk factors such as neurovascular invasion.

2.2. Large-panel DNA sequencing and data analysis

Tissue DNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tumor tissues using QIAamp DNA FFPE tissue kit (Qiagen, Hilden, Germany), following the manufacturer’s standard protocol. Capture-based targeted sequencing was performed with either a 520-gene panel (OncoScreen PlusTM, Burning Rock Biotech, Guangzhou, China)¹¹ or a 1021-gene panel (Geneplus-Beijing, Beijing, China).¹² For integrative genomic analyses across different sequencing platforms, we included only the genes shared by both the 520-gene and 1021-gene panels. Data analyses, including variants calling, interpretation, copy number variation, tumor mutational burden (TMB) estimation, and microsatellite instability (MSI) status assessment, were performed using standardized pipelines based on previously described methods.¹³

2.3. Multiplex immunohistochemical assay and image analysis

Multiplex immunohistochemical staining was performed using the PANO 7-plex IHC kit (Cat. 0004100100, Panovue, China) to identify specific cell subsets within the TME. Two distinct microenvironment analysis panels (MAP) were employed, each using a different set of primary antibodies. Panel 1 included PD-L1 (clone E1L3N, CST13684, CST), PD-1 (clone EH33, CST43248, CST), CD3 (clone SP7, Ab16669, Abcam), CD8 (clone C8/144B, CST70306, CST), and Pan-CK (clone C11, CST4545, CST). Panel 2 included CD68 (clone BP6036, BX50031, Biolynx), CD163 (clone D6U1J, CST93498, CST), CD56 (clone 123C3, CST3576, CST), and Pan-CK (clone C11, CST4545, CST). Following the primary antibody application, horseradish peroxidase-conjugated secondary antibodies, tyramide signal amplification (TSA), microwave treatment, and DAPI (Sigma-Aldrich) counterstaining were performed. Fluorescent whole-slide scanning was conducted using the Olympus VS200 system (Olympus, Germany) and the Mantra System (PerkinElmer, US) was used for analysis. The ratios of stained cells (PD-1⁺, PD-L1⁺, CD56⁺, CD3⁺, CD8⁺, CD163⁻CD68⁺, and CD163⁺CD68⁺) to total cells (DAPI-stained nuclei) were calculated in both intratumoral and stromal regions. TILs ratios in the intratumoral and stromal regions were expressed as iTILs and sTILs, respectively, with TILs referring to CD3⁺ and CD8⁺ cells.

2.4. Statistical and survival analysis

All statistical analyses and visualizations were performed using R software (version 4.0.0), with a significance level set at a two-tailed P-value of < 0.05. Categorical variables were compared using the Chi-square or Fisher's exact test, and continuous variables by the Wilcoxon rank-sum or Kruskal-Wallis test. Survival differences between 37 SRCC and 128 AC cases with complete follow-up data were analyzed. To minimize selection bias, propensity score matching (PSM) was used to pair 27 SRCC with 54 AC cases based on age, sex, and TNM stage. Sensitivity analyses based on the propensity score-matched cohort, including Rosenbaum bounds and E-value estimations, were conducted to evaluate the robustness of the findings against potential unmeasured confounding. Detailed methods and results are provided in the Supplementary methods. Prognostic parameters were analyzed in 33 SRCC cases, and survival differences related to the TME were evaluated using Kaplan-Meier curves and the log-rank test. Cox proportional hazards regression models were used to assess associations with disease-free survival (DFS) and overall survival (OS).

3. Results

3.1. Aggressive clinicopathologic features of SRCC

A comparison of 37 SRCC and 172 AC cases of the rectum and sigmoid colon revealed that patients with SRCC were significantly younger, with a median age of 55 years compared to 60 years for AC (P = 0.044). The male-to-female ratios were similar in both groups (SRCC: 1.9:1, AC: 1.5:1). SRCC tumors were more likely to exceed 5.0 cm in diameter (P = 0.008). SRCC cases were diagnosed at more advanced TNM stages, with 5.4 %, 70.3 %, and 21.6 % in stages II, III, and IV, respectively, compared to 17.4 %, 80.8 %, and 1.7 % in AC (P < 0.001). Morphologically, SRCC more often presented as long segmental circumferential thickening of the bowel wall (37.8 % vs. 3.5 %, P < 0.001). Regarding distant metastases, SRCC had significantly higher rates of peritoneal (37.8 % vs. 2.3 %, P < 0.001) and bone (18.9 % vs. 2.9 %, P = 0.001) metastases compared to AC (Table 1).

Table 1.

Clinicopathological characteristics of AC and SRCC.

Characteristic	AC (n = 172)	SRCC (n = 37)	P value
Age, (median, IQR), years	60.0 [51.6, 66.0]	55.0 [44.0, 63.0]	0.044
Sex, n (%)			0.755
Female	68 (39.5)	13 (35.1)
Male	104 (60.5)	24 (64.9)
Tumor size, n (%), cm			0.008
≤5.0	136 (79.1)	21 (56.8)
>5.0	36 (20.9)	16 (43.2)
TNM stage, n (%)			<0.001
II	30 (17.4)	2 (5.4)
III	139 (80.8)	26 (70.3)
IV	3 (1.7)	8 (21.6)
Unknown	0 (0.0)	1 (2.7)
Differentiation, n (%)			<0.001
Well	9 (5.2)	0 (0.0)
Moderately	131 (76.2)	4 (10.8)
Poorly	12 (7.0)	23 (62.2)
Unknown	20 (11.6)	10 (27.0)
Clinicopathological risk factorsa, n (%)			1.000
Yes	109 (63.4)	24 (64.9)
No	63 (36.6)	13 (35.1)
Diffuse circumferential thickening of the bowel wall, n (%)			<0.001
Yes	6 (3.5)	14 (37.8)
No	165 (95.9)	23 (62.2)
Unknown	1 (0.6)	0 (0.0)
CEA, n (%)			0.548
Abnormal	61 (35.5)	14 (37.8)
Normal	107 (62.2)	21 (56.8)
Unknown	4 (2.3)	2 (5.4)
CA19-9, n (%)			0.343
Abnormal	22 (12.8)	7 (18.9)
Normal	146 (84.9)	28 (75.7)
Unknown	4 (2.3)	2 (5.4)
MSI status, n (%)			0.049
MSI-H	4 (2.3)	4 (10.8)
MSS	168 (97.7)	33 (89.2)
TMB, (median, IQR), mutations/Mb	6.00 (4.80, 8.64)	4.80 (3.84, 6.86)	0.018

Abbreviations: AC, conventional adenocarcinoma; CA199, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; CRC, colorectal cancer; DFS, disease free survival; FFPE, formalin-fixed, paraffin-embedded; ICIs, immune checkpoint inhibitors; ICs, Immune checkpoints; IRB, Institutional Review Board; IS, Immunoscore; MAP, microenvironment analysis panels; MSI, microsatellite instability; NGS, next-generation sequencing; OS, overall survival; PD-1, programmed cell death protein 1; PSM, propensity score matching; PUMCH, Peking Union Medical College Hospital; SRCC, signet ring cell carcinoma; TILs, tumor-infiltrating lymphocytes; TLSs, tertiary lymphoid structures; TMB, tumor mutational burden; TME, tumor microenvironment; TNM, tumor, node, metastasis; TRMs, resident memory T cells; TSA, tyramide signal amplification.

^aClinicopathological risk factors include involvement of circumferential resection margin, positive extramural vascular invasion, or presence of perirectal cancer nodules as indicated in magnetic resonance imaging, or evidence of neurovascular infiltration in histological examination.

The median follow-up times were 38.0 months for SRCC and 59.5 months for AC cases. SRCC had significantly lower 3-year DFS (37.1 %) and OS (54.3 %) rates compared to AC (DFS 76.3 %, OS 90.1 %) (both P < 0.0001) (Fig. 1A and B). To account for confounding factors, PSM analysis was conducted, adjusting for age, sex, and TNM stage, with stage IV cases excluded due to their higher prevalence in SRCC and extremely poor survival. Despite these adjustments, SRCC still showed significantly worse 3-year DFS (P = 0.015) and OS (P = 0.008) compared to AC (Fig. 1C and D), highlighting the inherently aggressive nature of this rare subtype. Sensitivity analyses confirmed that the survival disadvantage of SRCC remained robust, as supported by both Rosenbaum bounds and E-value calculations (see Supplementary methods for details).

Fig. 1.

Survival of SRCC vs. AC. The Kaplan-Meier curves comparing the DFS (A) and OS (B) between SRCC (n = 37) and AC (n = 128). After 1:2 matching PSM analysis, the Kaplan-Meier curves comparing the DFS (C) and OS (D) between SRCC (n = 27) and AC (n = 54). AC, conventional adenocarcinoma; DFS, disease-free survival; OS, overall survival; PSM, propensity score matching; SRCC, signet ring cell carcinoma.

3.2. Distinct genetic features of SRCC

To better understand the genetic features of SRCC distinctive to AC, we performed next-generation sequencing in 37 SRCC and 172 AC cases (Fig. 2A). Fig. 2B and C illustrate significant differences in the commonly mutant genes between SRCC and AC. TP53 was the most frequently detected mutation in both groups, with a prevalence of 70 % in SRCC and 82 % in AC. APC and KRAS mutations, common in AC, were significantly less frequent in SRCC (APC, 32 % vs. 74 %, P < 0.001; KRAS, 16 % vs. 42 %, P < 0.01). FBXW7 mutations were exclusive to the AC group (20 %). In contrast, mutations in SMAD4 (32 % vs. 14 %, P < 0.05), RNF43 (19 % vs. 5 %, P < 0.05), MYC (14 % vs. 1 %, P < 0.001), ARID2 (11 % vs. 1 %, P < 0.01), and NCOR1 (11 % vs. 2 %, P < 0.05) were significantly more frequent in SRCC. In addition, BCL2L11 mutations were found exclusively in SRCC (14 %). In SRCC, the incidence of dMMR/MSI-H was significantly higher compared to AC (10.8 % vs. 2.3 %, P = 0.049) (Fig. 2D). A BRAF V600E mutation was found in one SRCC case (2.7 %), which also exhibited MSI-H. In contrast, four AC cases (2.3 %) harbored BRAF V600E variations, with one of these showing MSI-H. The TMB was significantly lower in SRCC compared to AC (median, 4.80 vs. 6.00 Mut/Mb, P = 0.0015) (Fig. 2E). After excluding dMMR/MSI-H cases to reduce the influence of random mutations, mutated genes in both SRCC and AC were found to be enriched in similar signaling pathways, including WNT, PI3K-AKT, RAS, MAPK, P53, ErbB, and mTOR (Fig. 2F). Interestingly, mutations in SRCC were generally confined to a few specific genes within each pathway, while AC exhibited a broader range of altered genes (Fig. 2G), suggesting a subtype-specific mutation-driven mechanism.

Fig. 2.

Mutational landscape of SRCC vs. AC. (A) The oncolplot (right) represents the top 20 frequently mutant genes in 37 SRCC and 172 AC patients with cancer of rectum and sigmoid colon. The heatmap (left) shows the percentage of the mutated genes detected in each group. Red denotes a higher proportion, and blue a lower proportion. (B and C) Comparison of frequency of mutant genes between SRCC (red bars) and AC (violet bars) within the whole population (B) and MSS cases only (C) (^⁎, P < 0.05; ^⁎⁎, P < 0.01; ^⁎⁎⁎, P < 0.001; and ^⁎⁎⁎⁎, P < 0.0001; Mann–Whitney U test). (D) The distribution of MSS and MSI-H cases in SRCC vs. AC is analyzed by Fisher’s test. (E) Comparison of TMB as the continuous value between SRCC and AC case using Mann–Whitney U test. (F and G) Comparison of signaling pathways involved in MSS cases between SRCC vs. AC. The proportions of patients with mutant genes attributed to a signaling pathway to total number of patients is shown as (F). The proportions of mutant genes detected to all listed genes within each pathway in the sequence panel is shown as (G). AC, conventional adenocarcinoma; MSS, microsatellite stable; MSI-H, microsatellite instability-high, SRCC, signet ring cell carcinoma; TMB, tumor mutational burden.

In this study, all four dMMR/MSI-H cases among the 37 SRCC patients remained free of recurrence or metastasis, with both 3-year DFS and OS reaching 100 %. Kaplan–Meier analysis revealed significantly better DFS compared to MSS cases (P = 0.042; Supplementary Fig. 1A) and a trend toward improved OS (P = 0.087; Supplementary Fig. 1B). The infiltration level of PD-1⁺CD8⁺ TILs in SRCC showed no significant association with MSI status in either the intratumoral or stromal compartments (P = 0.310 and P = 0.580, respectively; Supplementary Fig. 1C and D), suggesting that these two immune-related features may act independently.

3.3. Significantly higher CD8⁺ TILs but lower PD-1⁺CD8⁺ TILs in SRCC

Multiplex immunohistochemical analysis was performed on untreated tissue samples from 33 SRCC and 60 AC cases to elucidate the immune landscape specific to SRCC. The expression of immune cell markers, including CD3 (T cells), CD8 (cytotoxic T cells), CD56 (NK cells), CD68/CD163 (M1/M2 macrophages), and pan-cytokeratin (tumor cells), PD-1, and PD-L1, was assessed in both intratumoral and stromal regions. Subsequently, we compared the differences in immune cell types, densities, and distributions between SRCC and AC.

As shown in the heatmap (Fig. 3A), SRCC overall exhibited notably lower levels of immune cell infiltration within both the intratumoral and stromal regions compared to AC. However, SRCC demonstrated a significantly higher infiltration of CD8⁺ sTILs (P = 0.00024) and CD3⁺ sTILs in the stromal region (P = 0.00321) (Fig. 3C). In the intratumoral region, SRCC also showed higher CD8⁺ iTILs infiltration (P = 0.041), while CD3⁺ iTILs infiltration was comparable to AC (P = 0.657) (Fig. 3B). Despite the presence of high levels of CD3⁺ and CD8⁺ TIL infiltration, SRCC had significantly lower intratumoral PD-1 co-expressing TILs compared to AC (PD-1⁺CD8⁺ iTILs, P = 0.0035; PD-1⁺CD3⁺ iTILs, P = 5.0e-05) (Fig. 3B), a feature also visually evident in the multiplex immunohistochemical images (Fig. 3D). In AC, CD8⁺ iTILs significantly correlated with PD-1⁺CD8⁺ iTILs, whereas no such correlation was observed in SRCC (Supplementary Fig. 2).

Fig. 3.

The immune landscape of SRCC vs. AC. Quantitative analysis of immune cell infiltration was performed by multiplex immunohistochemical imaging in 33 SRCC and 60 AC cases. The ratio of each subtype of immune cells was calculated as the proportion of positive cells to all the cells in the intratumoral (i) and stromal (s) regions, respectively. (A) The heatmap shows the abundance of immune cells, represented by a gradient color scheme ranging from red (high-level infiltration) to blue (low-level infiltration). (B and C) The boxplots showing the ratios of immune cells in the intratumoral (B) and stromal (C) regions were compared between SRCC (red bars) and AC (violet bars) by Wilcoxon rank-sum test. Wilcoxon rank-sum test were used to calculate P values. (D) The multiplex immunohistochemical images of representative cases of SRCC (upper row) and AC (lower row), respectively showing the tumor infiltration of PD-1⁺ cells, TILs, and composite panoramas of all stained cells. Magnification: ×200; Scale bar, 100 µm. Although the infiltration levels of CD8⁺ TILs and CD3⁺ TILs are high in SRCC, the infiltration of PD-1⁺CD8⁺ TILs and PD-1⁺CD3⁺ TILs is significantly lower compared to AC. AC, conventional adenocarcinoma; PD-1, programmed cell death protein 1; SRCC, signet ring cell carcinoma; TILs, tumor infiltrating lymphocytes.

Compared to AC, SRCC exhibited significantly lower infiltration of M1 macrophages in both intratumoral (P = 0.0035) and stromal (P = 5.4e-09) regions, along with reduced stromal infiltration of M2 macrophages (P = 1.5e-05), and a borderline decrease in intratumoral M2 macrophage infiltration (P = 0.0522). NK cell infiltration levels were comparable between the two groups. Additionally, the intratumoral expression of PD-L1 was significantly lower in SRCC than in AC (P = 0.0031) (Fig. 3B and C).

3.4. High PD-1⁺CD8⁺ iTILs predicted longer survival in SRCC

We further investigated the prognostic value of TME parameters in 33 SRCC cases, using the median infiltration levels as cutoffs. Compared to patients with low infiltration, those with high intratumoral or stromal PD-1⁺CD8⁺ TIL infiltration demonstrated significantly longer DFS (intratumoral region, P = 0.00024; stroma, P = 0.0041) and OS (intratumoral region, P = 0.011; stroma, P = 0.0044), as depicted by Kaplan-Meier curves. Additionally, high stromal infiltration of PD-1⁺CD3⁺ sTILs was associated with longer DFS (P = 0.032) and OS (P = 0.0022), while high intratumoral infiltration of PD1⁺CD3⁺ iTILs was associated with longer DFS (P = 0.022) but not OS (P = 0.17) (Fig. 4). Even after excluding stage IV cases with extremely poor prognosis, high intratumoral PD-1⁺CD8⁺ iTIL infiltration remained a significant predictor of longer DFS (P = 0.0028) and OS (P = 0.024) in stage II-III SRCC. Moreover, among patients with normal baseline CEA levels, high intratumoral PD-1⁺CD8⁺ iTIL infiltration still exhibited a trend toward improved DFS (P = 0.062) (Fig. 5). However, the infiltration levels of CD3⁺ T cells, CD8⁺ T cells, M1/M2 macrophages, NK cells, PD-1⁺ cells, and PD-L1⁺ cells were not significantly associated with prognosis in patients with SRCC (Supplementary Fig. 3).

Fig. 4.

High-level infiltrations of PD-1⁺CD8⁺ TILs and PD-1⁺CD3⁺ TILs predicted improved survival in SRCC. Kaplan-Meier curves illustrating DFS (A, C, E, G) and OS (B, D, F, H) based on the infiltration levels of PD-1⁺CD8⁺ TILs and PD-1⁺CD3⁺ TILs in the intratumoral (I) and stromal (S) regions, respectively. Survival curves for cases with high (blue lines) and low (red lines) infiltration levels were plotted respectively using the median value of the ratio of TILs to all the cells as the cut-off. Log-rank tests were used to calculate P values. DFS, disease-free survival; OS, overall survival; PD-1, programmed cell death protein 1; SRCC, signet ring cell carcinoma; TILs, tumor infiltrating lymphocytes.

Fig. 5.

High-level infiltrations of PD-1⁺CD8⁺ TILs in the intratumoral (I) region predicted improved survival in stage II-III SRCC and indicated a tendency towards improved DFS in patients with normal baseline CEA levels. Kaplan-Meier estimates of DFS (A) and OS (B) based on the intratumoral infiltration levels of PD-1⁺CD8⁺ TILs in stage II-III SRCC patients. Kaplan-Meier estimates of DFS (C) and OS (D) based on the intratumoral infiltration levels of PD-1⁺CD8⁺ TILs in patients with normal baseline CEA levels. Survival curves for cases with high (blue lines) and low (red lines) infiltration levels were plotted respectively using the median value of the ratio as the cut-off. Log-rank tests were used to calculate P values. CEA, carcinoembryonic antigen; DFS, disease-free survival; OS, overall survival; PD-1, programmed cell death protein 1; SRCC, signet ring cell carcinoma; TILs, tumor infiltrating lymphocytes.

3.5. Immune cell infiltration correlated with genetic features

We hypothesized that the unique immune infiltration patterns of SRCC might be associated with distinct genetic profiles. Therefore, we explored the relationships between immune cell infiltration and genetic mutations. In SRCC, RNF43 emerged as a general pro-immune factor, significantly associated with increased infiltration of most immune cells. Conversely, BCL2L11 and TCF7L2 were primarily related to reduced infiltration of immune components (Supplementary Fig. 4A). In AC, however, the RNF43 mutation did not exhibit pronounced pro-immunogenic features. Instead, FBXW7, a gene mutation observed exclusively in AC, presented as the most significant immune promoter.

Regarding TILs co-expressing PD-1, in SRCC, ARID2 mutation was positively correlated with the infiltration of PD-1⁺CD8⁺ iTILs, while variations in MYC and NCOR1 were linked to decreased infiltration of PD-1⁺CD3⁺ iTILs (Supplementary Fig. 4A and Supplementary Fig. 5). In AC, FBXW7 mutation correlated with increased PD-1⁺CD3⁺ TIL infiltration. SMAD4 mutation was linked to reduced infiltration of PD-1⁺CD3⁺ and PD-1⁺CD8⁺ sTILs, and TP53 mutation was associated with decreased infiltration of PD-1⁺CD3⁺ iTILs (Supplementary Fig. 4B and Supplementary Fig. 6).

4. Discussion

Given the lack of research on colorectal SRCC, an important subtype in urgent need of effective treatment, this study provides a comprehensive analysis of its clinicopathological, genetic, and immune infiltration characteristics, with detailed comparisons to AC.

Building on previous studies,^3,14,15 this study confirms the poor prognosis of colorectal SRCC through PSM analysis, further validated by sensitivity analyses. Notably, 37.8 % of SRCC cases showed characteristic long-segmental circumferential bowel wall thickening with stenosis, resembling linitis plastica carcinoma, a distinct and challenging entity first described by Dixon and Stevens.¹⁶

Through NGS, we identified unique genetic features in SRCC, including fewer somatic variations and significantly lower mutation frequencies in common CRC driver genes such as APC, KRAS, and PIK3CA.^3,17, 18, 19, 20 While SRCC shares altered signaling pathways with AC, it may arise primarily from variations in a few SRCC-specific genes. For example, WNT pathway alterations in SRCC are more commonly driven by RNF43 mutations rather than the classical APC mutations, with RNF43 and APC mutations being mutually exclusive. To further explore the downstream functional differences between the two subtypes, we performed β-catenin immunohistochemistry on six SRCC samples harboring RNF43 mutations. All showed 10–60 % nuclear positivity, indicative of moderate activation of the canonical WNT signaling pathway. In contrast, APC mutations typically result in strong WNT pathway activation, highlighting potential functional divergence in WNT regulation between the two subtypes, which is consistent with previous reports.¹⁸ In addition, our results confirm previous findings^19,21 that SRCC has a higher proportion of MSI-H tumors than AC. However, unlike Western studies, MSI-H cases in this study showed no significant association with BRAF V600E mutations, possibly due to racial differences in the Asian population^20,22 and the exclusion of right-sided colon SRCC cases.

The characteristics of tumor immune infiltration hold important prognostic implications in CRC. Research indicates that TILs, particularly cytotoxic T cells, are powerful prognostic indicators, surpassing TNM staging.23, 24, 25 However, in this study, we observed a notable increase in CD8⁺ TIL infiltration in poorly prognosed SRCC, especially in its stromal regions, while the levels of TILs expressing PD-1 (both PD-1⁺CD8⁺ and PD-1⁺CD3⁺ TILs) were significantly lower than those in AC. Interestingly, our results also demonstrated significant heterogeneity in PD-1⁺CD8⁺ and PD-1⁺CD3⁺ TIL infiltration within SRCC, and higher infiltration of these TILs was significantly correlated with long-term survival, independent of MSI status. In contrast, the infiltration of CD8⁺ and CD3⁺ TILs were not predictive of survival in SRCC. Studies have shown that tumors often harbor a subset of non-tumor-specific CD8⁺ TILs, commonly termed bystander CD8⁺ T cells, within the TME. These cells are recruited via inflammatory chemokine gradients (e.g., CXCR3/CCR5 pathways) independent of tumor antigen recognition. Unlike tumor-specific T cells, bystander CD8⁺ T cells lack clonal expansion and typically express low level of PD-1, thereby diluting the functional cytotoxic pool of TILs.^26,27 Their accumulation may give rise to so-called “pseudo-T cell-inflamed” tumors.²⁸ In a single-cell multi-omics study, Borràs et al.²⁷ observed that CD8⁺ T cells infiltrating MSS CRC were more likely to exhibit non-tumor-specific or bystander-like profiles compared to those in MSI-H tumors. As a subtype recognized for its immunosuppressive properties, we infer that colorectal SRCC may contain a substantial proportion of non-tumor-specific T cells in its TME, potentially linked to the subtype’s characteristic non-tumor-specific inflammation.

PD-1-expressing CD8⁺ T cells were histologically regarded as exhausted and associated with poor prognosis.^9,10,29 However, accumulating evidence suggests that PD-1⁺CD8⁺ TILs do not necessarily represent dysfunctional cells; instead, many retain effector function and correlate with favorable prognosis in various cancers.^7,8,29, 30, 31, 32, 33 A TCGA-based study in 2020 reported that increased immune checkpoint expression in CRC correlated with higher TIL load, mutation rates, and improved survival.³⁴ Recently, Meyiah et al.⁶ analyzed tissue samples from 45 treatment-naive CRC patients and found that higher infiltration of PD-1⁺CD8⁺ TILs tended to link to longer DFS. Mechanistically, Thommen et al.³⁰ proposed that PD-1 expression may accurately identify a subset of clonally expanded, tumor-specific CD8⁺ TILs that exhibit high tumor-recognition capacity and retain multifunctionality, including the ability to secrete CXCL13 and recruit immune cells to form tertiary lymphoid structures (TLSs). Moreover, studies suggest that co-expression of PD-1 with CD39 and CD103 identifies truly tumor-reactive CD8⁺ T cells or tissue-resident memory T cells (TRMs) with enhanced anti-tumor activity, offering a more accurate predictor of favorable prognosis than total CD8⁺ T cell counts.35, 36, 37 Nonetheless, functional heterogeneity persists within the PD-1⁺CD8⁺ TIL population, warranting further research to delineate tumor-specific subsets with distinct immunological properties.

Upon analyzing immune cell infiltration, we further explored its association with genetic alterations to uncover potential mechanisms. Notably, alterations in MYC, NCOR1, and BCL2L11 were linked to significantly reduced immune cell infiltration, while RNF43 and ARID2 mutations were associated with higher MSI-H rates and increased TIL infiltration. Literature suggests that tumors with MYC alterations can dysregulate the TME and evade the host immune response. Targeting MYC may, therefore, simultaneously inhibit cancer cells and restore anti-tumor immunity.^38,39 In this study, the frequently mutated RNF43 in SRCC was associated with increased immune cell infiltration, likely due to the relatively high proportion of MSI-H cases (two out of seven). Previous studies have linked RNF43 mutations to MSI-H status in CRC.^40,41 Moreover, the pathogenicity of RNF43 may vary with its mutational pattern, leading to different effects on TME characteristics,⁴² which could explain the differences in gene-immune cell infiltration correlations between SRCC and AC. Furthermore, ARID2 alterations have been reported to promote intratumoral CD8⁺ T cell infiltration in melanoma mouse models, with ARID2 knockout sensitizing tumors to immune checkpoint inhibitors (ICIs).⁴³ These findings suggest that certain gene alterations may influence tumor-immune interactions and underlie the distinct immunogenomic profiles between SRCC and AC. Further pathway-level investigations integrating transcriptomic and functional omics data may help clarify how these genomic differences regulate key signaling pathways that shape immune phenotypes and tumor behavior.

A natural focus of interest lies in the response of MSI-H SRCC to immunotherapy. While MSI-H CRC typically demonstrate robust responses to ICIs, recent studies suggest that SRCC histology may attenuate this benefit. According to two independent studies, approximately half of patients with MSI-H colorectal SRCC exhibited poor responses to ICIs.^44,45 This resistance may be attributed to mucin-mediated stromal barriers and immune exclusion mechanisms characteristic of SRCC. Notably, in the NICHE trial,⁴⁶ PD-1⁺CD8⁺ TIL infiltration was identified as a potential predictor of ICI response in the pMMR subgroup, independent of MSI status. These findings suggest that functionally distinct TIL subsets—particularly PD-1⁺CD8⁺ T cells—may provide more accurate predictive value for ICI responsiveness than MSI status alone. Although our study did not involve immunotherapy intervention, we observed that intratumoral PD-1⁺CD8⁺ TILs, rather than total CD8⁺ TIL infiltration, were strongly associated with a favorable prognosis in SRCC. Based on this, we hypothesize that PD-1⁺CD8⁺ TIL density may serve as a potential biomarker for predicting ICI response in colorectal SRCC, independent of MSI status. This hypothesis warrants further investigation and prospective validation in immunotherapy-treated SRCC cohorts.

Given the rarity of SRCC, this study is limited by its observational nature and relatively small sample size. We explicitly acknowledge the limitations related to small-sample statistical power, the potential for overestimated effect sizes, and the need for external validation. Further studies with larger cohorts and investigations into the molecular characteristics of SRCC in the context of immunotherapy are needed to validate our findings and elucidate the underlying mechanisms, potentially offering insights to inform future clinical strategies, including immunotherapy.

5. Conclusions

Although rare, SRCC is the most aggressive subtype of CRC, exhibiting distinct genetic and immunological features. It is characterized by a “pseudo-T cell-inflamed” TME, marked by increased CD8⁺ TIL infiltration but reduced PD-1⁺CD8⁺ TIL density. Notably, PD-1⁺CD8⁺ TILs are significantly associated with improved survival in SRCC. These findings deepened our understanding of the immunogenomic landscape of colorectal SRCC and identifed PD-1⁺CD8⁺ TILs as a potential prognostic and therapeutic biomarker, supporting refined risk stratification and personalized therapeutic strategies for this challenging disease.

Author contributions

Jiaolin Zhou: study design, data integrity and accuracy. Jiaolin Zhou, Guole Lin and Huanwen Wu: research direction. Yang An, Lin Cong, Chentong Wang, Xinxin Mao, Bo Chen, and Han Chen: patient recruitment, clinicopathological information collection, and specimen preparation. Lan Su and Yang An: multiplex immunohistochemical assay analysis. Lan Su, Yaping Xu, and Yuhua Gong: NGS assays. Yang An and Lan Su: data analysis. Lan Su: figures and charts. Yang An, Jiaolin Zhou, and Lan Su: original manuscript writing. Jiaolin Zhou, Huanwen Wu, and Guole Lin: manuscript revision.

Ethical approval

The study was approved by the Institutional Review Board of PUMCH (approval number: I-23PJ672). All patients provided written informed consent for the use of their clinical data and biological samples.

Funding

This study was supported by CAMS Innovation Fund for Medical Sciences (CIFMS) (grant number: 2022-I2M-C&T-B-032), and National High Level Hospital Clinical Research Funding (grant number: 2022-PUMCH—C-005).

Declaration of competing interest

The authors declare no known competing financial interests or personal relationships that could have influenced the work. The companies affiliated with some authors provided no funding and played no role in study design, data interpretation, or manuscript preparation.