High KRT17 expression in tumour budding indicates immunologically ‘hot’ tumour budding and predicts good survival in patients with colorectal cancer

Abstract

Objectives

Emerging evidence has demonstrated that tumour budding (TB) is negatively associated with T-lymphocyte infiltration in CRC. Despite extensive research, the molecular characteristics of immunologically ‘hot’ TB remain poorly understood.

Methods

We quantified the number of TB by haematoxylin–eosin (H&E) sections and the densities of CD3⁺ and CD8⁺ T-lymphocytes by immunohistochemistry in a CRC cohort of 351 cases who underwent curative resection. We analysed the differential expression and T-lymphocyte infiltration score of 37 human epithelial keratins in CRC using RNA sequencing from the TCGA dataset. In 278 TB-positive cases, KRT17 expression was evaluated in tumour centre (TC) and TB with a staining score. Patient demographic, clinicopathological features and survival rates were analysed.

Results

In a CRC cohort of 351 cases, low-grade TB was associated with high CD3⁺ and CD8⁺ T-cell densities in the invasive margin (IM) but not in the TC. Of 37 human epithelial keratins, only KRT17 expression in TB had an apparent association with TB-grade and T-lymphocyte infiltration. In 278 TB-positive cases, high KRT17 expression in TB (KRT17TB) was negatively associated with low-grade TB and positively associated with high CD3⁺ and CD8⁺ T-cell densities in IM. High KRT17TB predicted early tumour grade, absence of lymph node metastasis and absence of tumour deposits. Additionally, patients with high KRT17TB had good overall survival and disease-free survival. Notably, low KRT17TB can specifically identify those patients with a poor prognosis among colorectal cancer patients with low TB and high T-lymphocyte infiltration.

Conclusions

KRT17 can be employed as a new indicator for distinguishing different immunological TBs.

Introduction

Colorectal cancer (CRC) is the fourth most frequent malignancy, with the second highest fatality rate.¹ Patients with metastatic CRC have a median overall survival of less than 2 years.² Tumour budding (TB), characterised by the presence of a single cell or a cluster of up to four cancer cells at the invasive margin (IM),³ is associated with lymphovascular invasion, nodal metastasis and distant metastasis.⁴ Moreover, TB stands as an independent adverse prognostic factor in the context of CRC.^5,6 Current findings have shown an inverse correlation between T-lymphocyte densities in CRC and TB at invasive fronts.^7,8 Low-grade TB characterised by extensive T-lymphocyte infiltration is seemingly linked to favourable survival outcomes in CRC.⁹ Nevertheless, the molecular features defining immunologically ‘hot’ TB remain predominantly undisclosed, highlighting the crucial need for identifying a distinctive marker to assess the invasiveness of such tumours.

Keratin represents a vital member of the intermediate filament family, playing an indispensable role in cellular structure and function. It is widely recognised that keratin serves as a distinctive component within the cytoskeleton of epithelial cells, playing a pivotal role in various cellular functions, including migration, differentiation and proliferation.^10,11 In the context of tumours, keratin plays diverse roles, encompassing functions such as epithelial-mesenchymal transition (EMT) and modulation of the tumour immune microenvironment.^12,13 Moreover, keratin is utilised as a diagnostic marker in the identification and classification of cancers originating from epithelial tissues.^14–16 Notably, keratin is frequently used to assess TB,^17–19 but there is a lack of research investigating the varied expression of keratin in TB among tumours characterised by different TB grades or distinct immunological states.

Within this investigation, a screening of 37 human epithelial keratins unveiled the pivotal involvement of KRT17 in the microinvasive processes of CRC. Within a CRC cohort consisting of 278 cases positive for TB, we observed a decline in KRT17 expression within tumour budding (KRT17TB) as the TB grade advanced. However, no such correlation was noted in the expression of KRT17 within the tumour centre (TC). Furthermore, KRT17TB exhibited a positive association with T-lymphocyte densities at the IM. Consequently, the discovery of elevated KRT17TB emerges as a novel indicator for immunologically ‘hot’ TB in CRC, holding immense significance in unravelling the intricate relationship between tumour microinvasion and the host immune system. In this context, to the best of our knowledge, we present the inaugural demonstration of the molecular features characterising immunologically ‘hot’ TB in CRC.

Results High KRT17TB is identified as a marker for immunologically ‘hot’ TB

Despite the assertion of an inverse correlation between TB levels and heightened lymphocyte infiltration,^7,20 the existence of a relationship between TB and the spatial distribution of T-lymphocyte infiltration remains unexplored. In order to substantiate the potential correlation between TB and T-lymphocyte infiltration in CRC, we conducted a comprehensive analysis of the extent of TB and T-lymphocyte infiltration in the tumour tissues of 351 CRC cases. Firstly, haematoxylin–eosin (H&E) sections from CRC cases were employed to assess the TB, while immunohistochemistry (IHC) was employed to scrutinise the presence of T lymphocytes (Figure 1a). Following the grading recommendations established by the International Tumour Budding Consensus Conference (ITBCC) in 2016,²¹ the entire cohort of CRC cases was categorised into three TB grades: low, intermediate and high TB. The level of T-lymphocyte infiltration was determined by CD3⁺ and CD8⁺ cells. To further analyse different positions in tumours, two tumour regions were examined, including the TC and the IM (Figure 1a). The findings revealed that T-lymphocyte levels in the TC displayed no significant association with TB grade, whereas T-lymphocyte levels in the IM exhibited a clear correlation (Figure 1b).

Figure 1. High KRT17TB is identified as a marker for immunologically ‘hot’ TBs. (a) Representative images of TB, CD3+ and CD8+ tissues of CRC patients. Scale bars: 100 μm (left), 20 μm (middle, right). (b) Quantification of CD3+ and CD8+ T lymphocytes in all 351 human CRC samples. n = 351. (c) Schematic showing the screening schedule to find out the candidate keratins for immunologically ‘hot’ TB. (d) Correlation between the four candidate keratin expression in TB and the number of TB, CD3+ density and CD8+ density in 20 CRC cases by IHC. n = 20. CD3IM, the CD3+ density in invasive margin; CD3TC, the CD3+ density in tumour centre; CD8IM, the CD8+ density in invasive margin; CD8TC, the CD8+ density in tumour centre; TBN, the number of tumours budding. Values are represented as the mean ± SD. *P [less than] 0.05, ns (no significance), by the two-tailed Student's t-test (d).

In order to identify the pivotal keratins influencing the immunological status of TB, a meticulously planned screening schedule was devised (Figure 1c). Firstly, the Cancer Genome Atlas (TCGA) database was utilised to ascertain the expression profiles of 37 human epithelial keratins. Out of the 37 keratins, 20 exhibited differential expression in tumour tissues when compared to normal tissue (Supplementary figure 1). Because lymphocyte signature and Th1:Th2 cell ratio were used as two important indicators to distinguish immune infiltration state²² and the role of Th2 cells in the infiltration of cytotoxic lymphocytes in CRC was not clear,^23,24 we explored the link between keratin expression and lymphocyte infiltration in the TCGA cohort by lymphocyte infiltration signature score and Th1 cells. Based on the online website²⁵ (https://www.camoip.net), lymphocyte infiltration signature score and Th1 cells in different groups of keratin expression in CRC were shown (Supplementary figure 2). The high- and low-expression groups of keratins were divided by median values. The results indicated a close association between 8 out of the 20 keratins and cytotoxic lymphocyte infiltration in CRC (Supplementary figure 2). However, 4 out of the 8 keratins exhibited down-regulation in tumour tissues in comparison to normal tissues, with their expression levels being exceedingly low in both normal and tumour tissues (Supplementary figure 1). Consequently, the 4 down-regulated keratins were omitted, leaving us with a final selection of 4 candidate keratins: KRT7, KRT17, KRT18 and KRT23.

To further explore the connection between the 4 candidate keratins and immunologically ‘hot’ TB, we analysed TB, keratin expression and T-lymphocyte infiltration in a CRC cohort of 20 cases. Surprisingly, the expression levels of the 4 candidate keratins in the TC exhibited no significant association with TB levels and T-lymphocyte infiltration (Supplementary figure 3). However, KRT17 expression within TB (KRT17TB) displayed a discernible association with both TB levels and T-lymphocyte infiltration (Figure 1d). To further prove the relationship between KRT17TB and TBN or T-lymphocyte infiltration, a CRC cohort of 278 TB-positive cases was established. The results showed that KRT17TB was negatively correlated with TBN and positively correlated with T-lymphocyte infiltration (Supplementary figure 4). These data indicate that KRT17 expression in TB could potentially regulate the tumour microenvironment by influencing the formation of TB and T-lymphocyte infiltration in CRC.

Higher KRT17TB is correlated with more T-lymphocyte infiltration in IM

To further confirm the correlation between the KRT17 expression profile at different sites and the degree of TB, we divided 278 TB-positive cases in the CRC cohort into a high KRT17 expression group and a low KRT17 expression group according to the median value. The expression of KRT17 in the TC displayed no apparent association with TB, while KRT17 within tumour budding (KRT17TB) exhibited a clear correlation (Figure 2a). These results indicated the potential role of a distinct expression profile of TB on tumour microinvasion at the invasive front of the tumour. A subsequent investigation into KRT17TB and T-lymphocyte infiltration in IM revealed that tumours exhibiting elevated KRT17TB were concomitant with heightened T-lymphocyte infiltration in IM. Subdivision of the cohort into subgroups based on TB levels, with 140 cases classified as low TB and 138 as intermediate and high TB, revealed consistent findings (Figure 2b and c). These data suggest that there may be an interaction between KRT17TB and T-lymphocyte infiltration, influencing tumour progression by modulating the formation and development of TB.

Figure 2. Higher KRT17TB is correlated with more T-lymphocyte infiltration in IM. (a) Quantification of KRT17 expression in TC and in TB by IHC score in all TB-positive human CRC samples. n = 278. (b) Representative images of TB, KRT17-, CD3- and CD8-stained tissues of TB-positive CRC cases. (c) Quantification of CD3+ and CD8+ T lymphocytes in all TB-positive human CRC samples. ImmHi TB, Immediate and high TB. n = 278. Values are represented as the mean ± SD. *P [less than] 0.05, ***P [less than] 0.001, ****P [less than] 0.0001, ns (no significance), by one-way ANOVA (a) and the two-tailed Student's t-test (c).

Characterisation and clinical signification of KRT17 expression in TB

Subsequently, to further explore the interaction between KRT17TB and malignant characteristics, an analysis was conducted to assess the association between KRT17TB and clinicopathological features in CRC (Figure 3a). In comparison to patients with early tumour grade, those with advanced tumour grade exhibited lower levels of KRT17TB. Nevertheless, there was no significant correlation observed between KRT17TB and T grade. The expression of KRT17TB exhibited down-regulation in cases with N1/N2 grade compared to those with N0 grade. Cases of moderately differentiated adenocarcinoma displayed lower levels of KRT17TB in comparison to well-differentiated adenocarcinoma. Those patients with the presence of tumour deposits exhibited lower levels of KRT17TB compared to those without tumour deposits. Surprisingly, there were no significant differences noted in the levels of KRT17TB with regard to the presence of lymphovascular invasion and perineural invasion. Our observations revealed that there were no variations in the levels of KRT17TB based on different KRAS statuses.

Figure 3. Characterisation and clinical signification of KRT17 expression in TB. (a) Relationship between KRT17TB and clinicopathological parameters of patients with TB-positive CRC. n = 278, except for n = 195 in KRAS status. (b) Kaplan–Meier survival curve of CRC patients layered by the KRT17 expression in TB in tumour tissue sections. All cases, n = 278; low TB cases, n = 140; intermediate and high TB cases, n = 138. P, Poor-differentiated adenocarcinoma; M, Moderate-differentiated adenocarcinoma; W, Well-differentiated adenocarcinoma; Mu, Mucinous adenocarcinoma. Values are represented as the mean ± SD. *P [less than] 0.05, **P [less than] 0.01, ***P [less than] 0.001, ns (no significance), by the two-tailed Student's t-test (a) and the log-rank test (b).

The Kaplan–Meier curves indicated a correlation between elevated KRT17TB and favourable outcomes in terms of overall survival (the log-rank test, P = 0.0002; hazard ratio, 0.3654; 95% CI [0.2146–0.6222]) and disease-free survival (the log-rank test, P < 0.0001; hazard ratio, 0.3266; 95% CI [0.2143–0.4977]) in all CRC cases (Figure 3b). Stratification of the cohort into patients with low TB (n = 140) or intermediate and high TB cases (n = 138) showed that higher KRT17TB was associated with overall survival in low TB cases (the log-rank test, P = 0.0048; hazard ratio, 0.2415; 95% CI [0.0907–0.6477]), whereas no significant association was observed in intermediate and high TB cases (the log-rank test, P > 0.05). However, high KRT17TB had less recurrence regardless of TB grades (low TB: the log-rank test, P < 0.0001; hazard ratio, 0.2018; 95% CI [0.1068–0.4162]; intermediate and high TB: the log-rank test, P = 0.0224; hazard ratio, 0.5046; 95% CI [0.2805–0.9075]). These results suggest that an interaction between KRT17TB and TB grade affects the prognosis of patients with CRC.

Association of KRT17TB and CD3IM with clinicopathological features

To further explore the association between KRT17TB and CD3IM with clinicopathological features, all cases were categorised into four groups: KRT17TB^lowCD3IM^low, KRT17TB^highCD3IM^low, KRT17TB^lowCD3IM^high and KRT17TB^highCD3IM^high (Figure 4a). Observations revealed that patients with grade III tumours had a larger proportion of KRT17TB^lowCD3IM^low than those with grade I/II, while those with advanced tumours had a smaller proportion of KRT17TB^highCD3IM^high. No significant differences were observed in the KRT17TB on different T grades. In comparison to patients with N0, those with N1 or N2 had a higher fraction of KRT17TB^lowCD3IM^low and a lower percentage of KRT17TB^highCD3IM^high. Based on differentiated level and histological type, moderate-differentiated tumours unexpectedly had the highest KRT17TB^lowCD3IM^low percentage and the lowest KRT17TB^highCD3IM^high proportion. However, the relationship between the two makers and the local histological invasion feature demonstrated that there were no significant differences in the four groups based on the absence or presence of tumour deposits, lymphocvascular invasion and perineural invasion. Patients with mutant KRAS tumours had a higher proportion of KRT17TB^lowCD3IM^low and a lower fraction of KRT17TB^highCD3IM^high.

Figure 4. Association between KRT17TB and CD3IM with clinicopathological features. (a) Relationship between four indicated groups and clinicopathological parameters of patients with TB-positive CRC. Patients were divided into four groups based on the median KRT17TB and CD3IM: KRT17TBlowCD3IMlow, KRT17TBhighCD3IMlow, KRT17TBlowCD3IMhigh and KRT17TBhighCD3IMhigh. n = 278, except for n = 195 in KRAS status. (b) Kaplan–Meier survival curve of CRC patients layered by the KRT17 expression in TB or the low CD3-positive lymphocytes in IM in tumour tissue sections. All cases, n = 139; Low TB cases, n = 113; Intermediate and high TB cases, n = 26. (c) Kaplan–Meier survival curve of CRC patients layered by the KRT17 expression in TB or the high CD3-positive lymphocytes in IM in tumour tissue sections. All cases, n = 139; low TB cases, n = 59; intermediate and high TB cases, n = 80. *P [less than] 0.05, **P [less than] 0.01, ns (no significance), by the Chi-squared test (a) and the log-rank test (b, c).

To further explore the key factors that affect prognosis, stratified analyses of the cohort were conducted (Figure 4b and c). In the low CD3⁺ T-cell density subgroup, elevated KRT17TB was not associated with good overall survival (the log-rank test, P > 0.05). Stratification of the cohort into patients with different TB grades found a similar result (the log-rank test, P > 0.05). However, on the one hand, higher KRT17TB was apparently associated with better disease-free survival (the log-rank test, P = 0.0197; hazard ratio, 0.4673; 95% CI [0.2465–0.8859]). Subgroup analyses suggested that patients with high KRT17TB had good survival in low TB-grade cases (the log-rank test, P = 0.0129; hazard ratio, 0.2865; 95% CI [0.1069–0.7674]), but not in intermediate and high-TB cases (the log-rank test, P > 0.05). On the other hand, among patients with high CD3⁺ T-cell density in IM, those with high KRT17 had better overall survival (the log-rank test, P = 0.0179; hazard ratio, 0.3186; 95% CI [0.1236–0.8211]) and disease-free survival (the log-rank test, P < 0.0001; hazard ratio, 0.222; 95% CI [0.111–0.4441]). Stratification of the cohort into patients with different TB grades found results similar to those with low CD3⁺ T-cell densities.

Association of KRT17TB and CD8IM with clinicopathological features

Based on the different expression levels of KRT17TB and cytotoxic T-lymphocytes in IM, the cohort was categorised into four groups (Figure 5a). Clinicopathological results were similar to CD3IM-related analysis, with the exception of RAS status. No significant differences were observed between patients with mutant KRAS tumours and those with wild-type. Stratification of the cohort into patients with low CD8⁺ T-cell density in IM (n = 141) or high CD8⁺ T-cell density in IM (n = 137) found that higher KRT17TB was associated with better overall survival (low CD8⁺ T-cell density in IM: the log-rank test, P = 0.0172; hazard ratio, 0.3844; 95% CI [0.1751–0.844]; high CD8⁺ T-cell density in IM: the log-rank test, P = 0.0044; hazard ratio, 0.325; 95% CI [0.1499–0.7047]) and disease-free survival (low CD8⁺ T-cell density in IM: the log-rank test, P = 0.0018; hazard ratio, 0.3819; 95% CI [0.209–0.698]; high CD8⁺ T-cell density in IM: the log-rank test, P < 0.0001; hazard ratio, 0.2725; 95% CI [0.1435–0.5175]) (Figure 5b and c). According to different TB grades, subgroup analyses found similar results in the low-TB grade cases (low CD8⁺ T-cell density in IM: OS, the log-rank test, P = 0.0446; hazard ratio, 0.1838; 95% CI [0.0352–0.9602]; DFS, the log-rank test, P = 0.0032; hazard ratio, 0.2294; 95% CI [0.0862–0.6107]; high CD8⁺ T-cell density in IM: OS, the log-rank test, P = 0.0091; hazard ratio, 0.1694; 95% CI [0.0446–0.6431]; DFS, the log-rank test, P = 0.0002; hazard ratio, 0.1379; 95% CI [0.0485–0.3919]), but not in the intermediate and high-TB cases (low CD8⁺ T-cell density in IM or high CD8⁺ T-cell density in IM: the log-rank test, P > 0.05). Collectively, these data indicated a crucial role for KRT17TB in the relationship between TB and T-lymphocyte infiltration in IM.

Figure 5. Association between KRT17TB and CD8IM with clinicopathological features. (a) Relationship between four indicated groups and clinicopathological parameters of patients with TB-positive CRC. Patients were divided into four groups based on the median KRT17TB and CD8IM: KRT17TBlowCD8IMlow, KRT17TBhighCD8IMlow, KRT17TBlowCD8IMhigh and KRT17TBhighCD8IMhigh. n = 278, except for n = 195 in KRAS status. (b) Kaplan–Meier survival curve of CRC patients layered by the KRT17 expression in TB or the low CD8-positive lymphocytes in IM in tumour tissue sections. All cases, n = 141; Low TB cases, n = 105; Intermediate and high TB cases, n = 36. (c) Kaplan–Meier survival curve of CRC patients layered by the KRT17 expression in TB or the high CD8-positive lymphocytes in IM in tumour tissue sections. All cases, n = 137; low TB cases, n = 67; intermediate and high TB cases, n = 70. *P [less than] 0.05, **P [less than] 0.01, ns (no significance), by the Chi-squared test (a) and the log-rank test (b, c).

Discussion

Emerging evidence strongly suggests that immunologically ‘hot’ TB serves as a critical marker for the survival of CRC patients.^20,26,27 Nevertheless, the molecular subtype of TB remains unclear for distinguishing among different immunological statuses in CRC. The purpose of this study was to identify a marker capable of distinguishing TB based on different immune statuses. Our findings revealed that elevated levels of KRT17TB were inversely correlated with the grade of TB and positively associated with T-cell densities in the IM. Furthermore, individuals with elevated levels of KRT17TB exhibited a propensity towards early tumour grades and a lower incidence of lymphatic metastasis in CRC. CRC patients with elevated levels of KRT17TB were associated with improved survival outcomes. Despite the favourable prognostic indications associated with both low-grade TB and significant T-lymphocyte infiltration in CRC patients, our study identified that low levels of KRT17TB sensitively identify individuals at risk of relapse and with shorter survival in these populations. This study emphasises the clinical importance of the local immune microenvironment in the IM concerning tumour microinvasion. These findings may offer additional insights into the mechanism of tumour microinvasion, particularly for TB-positive tumours. To our knowledge, this is the first study to seek out a marker for distinguishing TB from different immune status.

Tumour budding has recently been shown to have a strong predictive value in patients with CRC.^28,29 It has long been considered that cells undergoing EMT contribute to TB.^30,31 Thus, an EMT-associated gene expression signature in TB, including repressors of E-cadherin and β-catenin, has also been investigated in the context of TB.³² Many proteins have been shown to be indicators of the formation of TB and survival.^33–35 Emerging studies have shown that the number of TB cases is negatively associated with T-lymphocyte infiltration.^7,20 Immunologically ‘hot’ TB, which refers to the TB with abundant T-lymphocyte infiltration, is associated with good survival. However, there is no study for an indicator to identify those immunologically ‘hot’ TB. Our findings demonstrated that KRT17 was highly expressed in low-grade TB, which is positively associated with T-lymphocyte infiltration. Patients with high KRT17TB have been shown good to have overall survival and less recurrence, which may provide evidence for strategies to screen out high-risk factors for neoadjuvant therapy in CRC.

Our study provides evidence for the pro-inflammatory role of KRT17. There is still controversy about the immune regulation of KRT17 in pathologic conditions. In dermatitis and skin tumourigenesis, KRT17 promotes the pro-inflammatory tumour microenvironment through multiple mechanisms, such as increasing chemokine expression and activating the NF-κB pathway.^36–38 However, Wang and colleagues have reported an immune-escape role of KRT17 in head and neck cancer.³⁹ The reason for this discrepancy remains unclear, but it may be because the function of KRT17 within the tumour microenvironment may differ depending on the cancer type. Our research increased new evidence for a pro-inflammatory role of KRT17 through a large-scale clinical cohort in CRC.

Importantly, we found that TB exhibits different expression patterns compared with TC and that TB has a close link with T-lymphocyte infiltration in IM rather than in TC, indicating that the interaction between tumours and T-lymphocytes at the invasive front might represent a microinvasion condition. Consistently, emerging evidence has proposed that TB exhibits different expression patterns of EMT-related markers, specifically increased mesenchymal-related markers and decreased epithelial-related markers.^32,40 Zhou and colleagues have confirmed that activation of FAK and Yes-associated proteins (YAP) by the interaction between LN-52 and integrin 1 enhanced the TB of CRC.⁴¹ Loss of MHC class I expression has been observed in TB and is linked to a poor prognosis because it can make it possible to evade the adaptive immune response.⁴² These studies emphasise the importance of different microenvironments for tumour microinvasion, and it is of great clinical value to explore potential markers. We found that KRT17 expression in TB represents not only the number of TB but also the level of T-lymphocyte infiltration in IM, indicating that KRT17 has extensive clinical prospects.

We have established an attacker-defender model to elucidate the intricate interactions between TB and the defences mounted by inflammatory cells against them.^7,17 TB serves as a hallmark of an aggressive disease phenotype, associated with an elevated risk of recurrence in the liver and lung, as well as poor overall survival in CRC patients.⁴³ The level of T-lymphocyte infiltration is predominantly considered a key defensive mechanism. Consequently, the integration of TB and T-cell densities enhances the prediction of nodal metastases and improves survival prognostication in CRC patients.^8,9 Deciphering the immunological escape mechanisms of TB and understanding the cues from the tumour stroma that foster budding holds the promise of introducing novel therapeutic avenues. Nevertheless, there remains a dearth of markers capable of representing the conditions in the attacker-defender model. Our discoveries furnish evidence for the outcomes at the invasive front, aiding doctors in designing effective treatment strategies.

It is crucial to interpret this study within the context of its limitations. The primary constraints of our study stem from its retrospective cohort design. However, we sought to mitigate the impact of tumour heterogeneity by enrolling a comprehensive cohort comprising 351 patients with CRC. It is important to note that our study's generalisability might be constrained as all cases originate from a single centre. Nevertheless, our study capitalises on the availability of comprehensive clinicopathological data and detailed prognosis records. Notably, the inclusion of only a limited number of patients with T1 carcinomas is attributed to the restricted patient pool and the amount of available cancer tissue.

In conclusion, our findings indicate that elevated KRT17TB is linked to low-grade TB and heightened T-lymphocyte infiltration. Patients exhibiting high KRT17 expression in TB tend to experience better prognoses compared to those with lower KRT17 expression in TB. Notably, among TB-positive patients, their survival is more closely associated with KRT17 expression in TB than with T-lymphocyte densities. To the best of our knowledge, this study represents the inaugural exploration of the molecular subtype of TB in relation to different immune statuses in CRC.

Methods Patients and tissue samples

The formalin-fixed, paraffin-embedded (FFPE) CRC tissue samples were obtained from 351 CRC patients who had received radical surgery at the Sixth Affiliated Hospital of Sun Yat-sen University. Clinical parameters were obtained from the patient's electronic medical records located in our hospital. The histomorphological characteristics were reviewed from the corresponding H&E-stained slides. All human tissue samples were obtained with written informed consent from donors, and all procedures were conducted with the permission of the Institutional Review Board of The Sixth Affiliated Hospital of Sun Yat-sen University.

Assessment of tumour budding

The H&E-stained slides were used to examine TB by two pathologists. TB was defined as the occurrence of a single cell or a cluster of up to four cancer cells at the IM, and graded as low (0–4 buds per 0.785 mm²), immediate (5–9 buds per 0.785 mm²) and high (≥ 10 buds per 0.785 mm²) based on the International Tumour Budding Consensus Conference (ITBCC) grading recommendations in 2016.²¹

Immunohistochemistry

The primary antibodies used here were anti-CD3 (Abcam, Cat#ab135372, 1:1000), anti-CD8 (Novus, Cat#NBP2-29475, 1:1000), anti-KRT7 (Abcam, Cat#ab68459, 1:1000), anti-KRT17 (Abcam, Cat#ab51056, 1:1000), anti-KRT18 (Abcam, Cat#ab133263, 1:1000) and anti-KRT23 (Abcam, Cat#ab156569, 1:100). Mouse/Rabbit Polymer Test System Universal Kit (ZS-bio, Beijing, China) was used to perform IHC according to its instructions. Briefly, paraffin-embedded tissues were deparaffinised with dimethylbenzene and citrate solution (Servicebio, Wuhan, China) was used for antigen retrieval. The tissues were incubated overnight at 4°C with the indicated primary antibodies. Then the tissues were incubated with enzyme-labelled anti-goat IgG polymer and incubated with 3, 3′-diaminobenzidine (DAB) complex. Finally, the tissues were counterstained with haematoxylin (ZS-bio, Beijing, China). The densities of CD3⁺ and CD8⁺ cells were calculated as the number per mm² of positive cells in two tumour regions: TC and the IM. The expression of four candidate keratins was quantified by the staining scores, which were based on the staining intensity and proportion of positively stained cells. Similarly, each keratin expression was determined on the TC and TB regions. The staining intensity was divided into the following 4 groups: grade 0 with no staining, grade 1 with weak staining, grade 2 with medium staining and grade 3 with strong staining. The proportion of positively stained cells was determined by the percentage of positive stained area using the following 4 groups: grade 0 = 0, grade 1 = 1–25%, grade 2 = 26–50%, grade 3 = 51–75% and grade 4 ≥ 75%. The staining scores were calculated using the following formula: staining scores = staining intensity × proportion of positively stained cells. Based on this method, the expression levels of four candidate keratins were determined by the staining scores on a scale from 0 to 12. The expression levels of CD3, CD8 and KRT17 were divided into low and high groups according to their median.

Statistical analysis

All data are presented as mean ± standard deviation (SD) unless otherwise noted. All analyses were performed with SPSS 25.0 (SPSS 25.0; SPSS Inc., Chicago, Illinois, USA). For continuous variables with normal distributions, a two-tailed Student's t-test or one-way ANOVA was employed to establish statistical significance, whereas either a Mann–Whitney U-test or a Kruskal–Wallis test was applied for skewed distributions. A Pearson's chi-squared test or Fisher's exact test was used to compare categorical variables when appropriate. For survival analysis, Kaplan–Meier plots and log-rank tests were employed. All P-values were two-sided, and statistical significance was defined as a P-value of 0.05.

Acknowledgments

This work is supported by the National Natural Science Foundation of China (82200569 and 82000515), Science and Technology Projects in Guangzhou (202206010062), Science and Technology Key Research and Development Plan Project of Guangzhou (China) (202103000072), China Postdoctoral Science Foundation (2021M703723), Guangdong Basic and Applied Basic Research Foundation (2022A1515012498, 2021A1515111011 and 2019A1515110043), Sun Yat-sen University Clinical Research 5010 Program (2016005), Shenzhen ‘San Ming Projects’ Research (Grant No.lc202002), the Program of Guangdong Provincial Clinical Research Center for Digestive Diseases (2020B1111170004) and National Key Clinical Discipline.

Author contributions

Wenfeng Liang: Data curation; investigation; methodology; software; visualization; writing – original draft. Haiqing Jie: Data curation; formal analysis; methodology; software. Hao Xie: Data curation; formal analysis; methodology; software. Yebohao Zhou: Data curation; methodology; software. Wenxin Li: Formal analysis; methodology; software; visualization. Liang Huang: Funding acquisition; investigation; methodology; validation; visualization. Zhenxing Liang: Data curation; methodology; software. Huashan Liu: Data curation; methodology; software. Xiaobin Zheng: Funding acquisition; validation; visualization. Ziwei Zeng: Conceptualization; resources; supervision; validation; visualization; writing – review and editing. Liang Kang: Conceptualization; funding acquisition; project administration; resources; supervision; validation; writing – review and editing.

Conflict of interest

The authors declare no conflict of interest.

Data availability statement

The authors declare that all data supporting the findings of this study are available within the paper and the Supporting Information and from the authors on request.

Ethics approval and consent to participate

All human tissue samples were obtained with written informed consent from donors, and all procedures were conducted with the permission of the Institutional Review Board of The Sixth Affiliated Hospital of Sun Yat-sen University. All procedures were carried out in accordance with the Declaration of Helsinki.