Background

Colorectal cancer (CRC) is one of the major malignancies in the world and the second-leading cause of cancer deaths [1], medical advances notwithstanding. The basis for curative therapy is surgical resection, but still around 40% of the patients die from metastatic disease. In addition to the TNM staging system, the tumour immune response has been found an important prognosticator in CRC, with a higher density of tumour-infiltrating lymphocytes predicting a better outcome [2]. This finding has led to a joint effort to introduce the Immunoscore, based on immunohistochemical (IHC) evaluation of T cell markers, into clinical practice [3]. T cells are known combatants in anti-tumour immunity and can inhibit tumour growth by direct killing (cytotoxic T cells) [4]. The activity of the cytotoxic T cells is, however, tightly regulated by the balance of other T cell subsets; the T helper 1 (Th1) and T helper 2 (Th2) cells, and the T regulatory cells, as well as the expression of checkpoint inhibitors, e.g., CTLA4 and PD1/PDCD1 [5].

Considering the well-established prognostic value of the tumour immune response in CRC, these patients should be promising candidates for immunotherapy. However, the clinical response to immune checkpoint inhibitors has so far been quite limited in patients with metastatic CRC, the strongest effect found in patients with microsatellite instable (MSI) tumours [6]. The Food and Drug Administration has recently approved Pembrolizumab and Nivolumab (PD1 inhibitors) for clinical use in the treatment of metastatic MSI CRC patients, but ongoing studies suggest a striking beneficial effect of immune checkpoint inhibitor therapy also in non-metastatic MSI CRC [7, 8].

Sporadic CRC is a heterogenous disease, and the heterogeneity is partly seen in its development through three major genetic pathways: The classical chromosomal instability adenoma to carcinoma pathway responsible for approximately 70% of sporadic CRCs, the MSI pathway caused by aberrations in mismatch repair genes found in around 15% of CRCs, and the epigenetic instability CpG island methylator phenotype pathway causing 30–40% of sporadic CRCs. These pathways are partly driven by mutations in tumour oncogenes and are not mutually exclusive. Some of the most common driver mutations found in sporadic CRC are activating mutations in KRAS accounting for 30–40% of all CRCs and activating mutations in BRAF which occur in 5–15 % of the tumours [9].

Many robust studies have linked immune cell infiltration to an immunogenic MSI/d-MMR tumour status [4, 10]. However, the Immunoscore has been shown to be a stronger predictor of patient survival than tumour MSI status, suggesting that there are other subgroups of CRC patients in whom immune cell infiltration has clinical value [11]. Indeed, studies by us and others have proposed that there are microsatellite stable (MSS) tumours with a similar tumour immune profile as MSI tumours, e.g., the POLE mutated CRCs [11,12,13,14]. In addition, the expression of immune checkpoint proteins such as PD1 and PD-L1/CD274 are not restricted to MSI tumours [15, 16]. More comprehensive studies on the distribution of immune cell subsets according to KRAS and BRAF mutation are rare. KRAS and BRAF mutations are mainly mutually exclusive [17]. BRAF mutation has been associated with MSI tumours, but is also found in MSS tumours [18]. BRAF-mutated tumours have also been linked to a higher immune cell infiltration and expression of immunotherapeutic targets [19], but the relative contribution of mutated BRAF and MSI to the immune response remains uncertain. The potential of extracting putative predictive information from KRAS and BRAF mutation analyses in immunotherapy would be attractive since these are often already included, in addition to MSI status, in the clinical assessment of CRC patients.

This project aims to better characterise the local immune response to CRC in association with molecular characteristics. In particular, the importance of KRAS and BRAF mutation on immune cell infiltration will be evaluated. An increased understanding of the distribution of the immune response in different molecular subgroups of CRC may lead to the identification of prognostic and predictive tools, as well as new targets for therapy.

Materials and methods

Study cohort of patients with CRC

A logistic system has been developed for collecting biological samples and data from CRC patients diagnosed at Umeå University Hospital, Umeå, Sweden. Since 2010, CRC patients are enroled in U-CAN (Uppsala-Umeå Comprehensive Cancer Consortium) [20], which longitudinally collects blood, tissue, faeces, radiological data, and clinical data from all enroled patients. The present study utilised retrospectively collected resected formalin-fixed paraffin-embedded (FFPE) primary tumour tissue specimens from a cohort of stage I–IV U-CAN patients with faecal samples collected between the years 2010–2014 (n = 257), for studies of microbial markers in relation to molecular subtypes, immune cell infiltration and prognosis [21]. The patient tumour tissue specimens of this cohort have been analysed for molecular changes including BRAFV600E mutation and KRAS mutations (codon 12 and 13), and MSI status, as previously described [22]. Also data on clinicopathological characteristics, and survival have been collected. From this cohort a tumour tissue microarray (TMA) was prepared in September 2020, from patients operated on in Umeå with available tumour tissue at that time (n = 151). The TMA blocks were constructed using a TMA GRAND Master instrument (3DHISTEC, Budapest, Hungary) by punching 1 mm cores from archived FFPE tissue samples. The TMAs used for this study, included two cores of tumour tissue taken at the tumour front and one core taken from the tumour centre for each patient.

For validation, retrospectively collected tumour specimens from CRC patients identified within a population-based cohort from northern Sweden, the Northern Sweden Health and Disease Study (NSHDS), were used. The NSHDS cohort has been described in detail previously [23]. Patients diagnosed between the years 2000–2016 with available tumour tissue for analyses of molecular traits and immune infiltration were included in this study (n = 608). Clinical patient characteristics can be found in Supplementary Table S1.

Multispectral imaging for in situ evaluations of immune cell subsets in CRC

We evaluated immune cell infiltration using the Vectra® system for multispectral quantitative automated pathology imaging (Akoya Biosciences). The multiplexed IHC staining was a modification of the OpalTM 7 Solid Tumour Immunology Kit (Akoya Biosciences). Tissue TMA slides were sequentially stained using antibodies against T-bet, CD8, CD20, FoxP3, CD68 and pan-Cytokeratin. The CD4 antibody used in the kit was exchanged by T-bet (clone H210, Santa Cruz Biotechnology Inc.). The CD8 antibody in the kit was exchanged by CD8 (clone 144b, DAKO). The final concentrations of antibodies and Opal dyes used can be found in Supplementary Table S2. The Prolong Diamond Antifade Mountant (ThermoFisher, Waltham, MA, USA) was used to mount slides. Multispectral imaging was performed using the VECTRA 3 Quantitative Pathology Imaging System (Akoya Biosciences) as previously described [24]. A spectral library was created and applied to the images for spectral unmixing together with an autofluorescence control using the inForm® software (Akoya Biosciences, Marlborough, USA). The signal intensity of each marker was adjusted to give a signal range of 5–30 ms at exposure times of 30–200 ms. For validation, multiplex images were compared to monoplex images and inspected for interference and cross-talk. The inForm® software (Akoya Biosciences) was further used to quantify the different immune cells in tissue segmented areas using machine learning algorithms, as previously described [24]. T-bet was used as a marker for T helper 1 cells, but has been reported to be expressed also in cytotoxic (CD8 positive) T cells [25]. In this study, CD8 positive cells generally showed no or low expression of T-bet and were classified as cytotoxic T cells. Remaining T-bet positive cells were defined as T helper 1 cells. Tissue was segmented into tumour epithelial area, stromal area, tumour debris and no tissue. Extracellular mucin was classified as tumour debris. Exclusions were, large part of or whole core lost, lack of tumour or stromal tissue, bad quality of staining, and heavy necrosis. Image analysis was supervised by two experienced pathologists, and the investigators were blinded from patient data during the evaluations. Cell segmentation summary data was collected for each TMA core as number of individual cell types per mm2 (tumour or stromal compartment). For patients with data on two TMA cores, data was given as an average number of cells per mm2. Data was successfully collected from tumour front of 139 patients (of which 33 had data from only one core), and from tumour centre of 137 patients. Infiltrating immune cells, as identified by the respective markers, were further divided into groups of high and low infiltration with the median number of infiltrating cells for all CRC cases used as cut-off. Patients not surgically resected for CRC and patients dying from post-operative complications within 90 days of surgery were excluded from survival analyses.

Evaluation of cytotoxic T cells by immunohistochemistry for validation

The evaluation of cytotoxic T cells (CD8+) by immunohistochemistry in the NSHDS validation cohort has been previously described [23]. In brief, CD8+ cells were stained on whole tissue sections and semi-quantitatively evaluated at the tumour front of CRC specimens (n = 608) ranging from score (1) no or sporadic, (2) moderate numbers, (3) abundant or (4) highly abundant. Patients with pre-operatively irradiated rectal tumours were excluded (n = 98), leaving in total 510 CRC tumour specimens for analyses.

Statistics

IBM SPSS Statistics v.28 (SPSS Inc.) was used for statistical analyses. The distribution of each variable and subgroup was assessed and appropriate statistical tests chosen accordingly. Comparisons of differences within continuous variables between groups were hence performed by non-parametric two-sided Mann–Whitney U tests or Kruskall-Wallis H tests. The χ2 test or Fischer´s exact test (when expected counts were less than 5) was used for comparisons of categorical variables. Cancer-specific survival was defined as death with disseminated or recurrent disease and estimated using Kaplan–Meier survival analyses. Follow-up time was from day of surgery until death or end of follow-up (October 2021 for U-CAN, May 2022 for NSHDS). Median follow-up time was 92.8 months for U-CAN and 94.5 months for NSHDS. Log-rank tests were used to estimate differences in 5-year cancer-specific survival between groups. The Cox Proportional-Hazards model was used for uni- and multivariable survival analyses. A P value < 0.05 was considered statistically significant.

Results

Distribution of different immune cells in tumour front and centre of CRC patients

We have set up a functional assay for multiplex immunohistochemistry and evaluation of CD8+ cytotoxic T cells, T-bet+ Th1 cells, FoxP3+ T regulatory cells, CD20+ B cells, CD68+ macrophages, and cytokeratin using the VECTRA system for multispectral imaging (Fig. 1). We have evaluated this immune cell infiltration panel in surgically resected stage I–IV patients from a CRC patient study cohort (n = 257). The different immune cells were evaluated in the stromal compartment and in the tumour compartment at the tumour front and in the tumour centre. The infiltrating immune cells were shown to be quite evenly distributed between the tumour front and the tumour centre, but with higher levels in the stromal area compared to the tumour epithelial area (Table 1). Data from the tumour epithelial area for FoxP3+ T regulatory cells and CD20+ B cells was not further analysed since these immune cells were rarely present within this compartment.

Fig. 1: Multispectral imaging of immune cell infiltration in tumour front and centre of CRC study patients.
figure 1

Shown are high-magnification areas (20x) of representative multiplex IHC stains from (a) a TMA (1 mm core) taken at the tumour front and (b) a TMA (1 mm core) taken at the tumour centre. Left panel displays multiplex IHC images after spectral unmixing, of cytokeratin (magenta), CD8+ cytoxic T cells (red), T-bet+ Th1 cells (white), FoxP3+ T regulatory cells (orange), CD20+ B cells (yellow), and CD68+ macrophages (cyan). Right panel displays the same multiplex IHC but with tumour tissue segmentation map of stromal area (blue), tumour epithelial area (magenta), tumour debris (yellow) and no tissue (brown), and the cell phenotypes from which data was collected highlighted using the above described colours.

Full size image
Table 1 Distribution of immune cells at the tumour front and centre in CRC tumours.
Full size table

Analyses of immune cell infiltration according to patient clinical and pathological characteristics

The strongest associations of immune cells to clinical and pathological parameters in study patients were found at the tumour front (Table 2). A slightly higher infiltration of immune cells was found in women, especially within the tumour epithelial area. Immune cell infiltration was also strongly linked to the right-sided colon cancers, except for B cells and macrophages. All immune cell types were found linked to tumour stage, with a gradual decrease in immune cell infiltration from stage I to stage IV tumours (Table 2). Pre-operative radiotherapy of rectal cancers was strongly associated with a decreased infiltration of immune cells. Thus, for further analyses, the pre-operatively irradiated rectal cancers were excluded. The clinical and pathological associations of immune cell infiltration in the tumour centre can be found in Supplementary Table S3.

Table 2 Associations of immune cells in tumour front to clinicopathological characteristics of CRC patients.
Full size table

Analyses of immune cell infiltration in molecular subgroups of CRC

Infiltrating immune cells at the tumour front of study patients further exhibited several significant associations with tumour molecular subtypes (Table 3). The MSI tumours were found to be significantly more infiltrated by cytotoxic T cells, Th1 cells, and macrophages, in validation of our analyses. We found contrasting associations of immune cell infiltration in KRAS-mutated and BRAF-mutated CRC tumours. While KRAS-mutated tumours showed less infiltration of immune cells compared to wild-type KRAS tumours, in particular cytotoxic T cells and Th1 cells, BRAF-mutated tumours showed a higher infiltration of these immune cells compared to wild-type BRAF tumours. No associations to KRAS or BRAF mutation was found for T regulatory cells at the tumour front. The associations of immune cells in the tumour centre to molecular characteristics can be found in Supplementary Table S4. Similar, but weaker associations were found in the tumour centre. However, in the tumour centre KRAS mutation was significantly associated with increased infiltration of regulatory T cells (Supplementary Table S4).

Table 3 Associations of immune cells in tumour front to molecular characteristics of CRC tumours.
Full size table

To investigate the relative contribution of BRAF mutation and MSI status on immune cell infiltration, we stratified the analysis of immune cell infiltration according to BRAF mutation by MSI status (Table 3). Even though limited by sample size, BRAF mutation was found to increase infiltration of cytotoxic T cells in both MSI and MSS tumours to similar levels. In contrast, Th1 cell infiltration appeared to be more linked to MSI status than BRAF mutation. However, BRAF mutation was still linked to increased infiltration of most analysed immune cells (with the exception of regulatory T cells) in the MSS tumours.

In validation analyses, tumour infiltration of cytotoxic T cells was evaluated in an independent cohort of CRC patients, the NSHDS (n = 608). Also here, a contrasting role on immune cell infiltration by KRAS and BRAF mutation was found. The proportion of KRAS mutated tumours was found to decrease with increasing score of infiltrating cytotoxic T cells, while the proportion of BRAF mutated tumours instead was found to increase (Table 4). The proportion of MSI tumours was also found to increase with increasing immune score (Table 4). However, even though MSI status was found highly associated with a high immune score, around half of the tumours with the highest immune score were defined as MSS (Table 4). In this cohort where infiltration of cytotoxic T cells was assessed on a semi-quantitative scale (score 1-4), immune cell infiltration by BRAF mutation was found mainly linked to MSI tumours (Table 4).

Table 4 Associations of cytotoxic T cells in tumour front to molecular characteristics of CRC tumours in the validation cohort.
Full size table

Analyses of immune cell infiltration in CRC prognosis

A positive prognostic effect of immune cell infiltration in CRC tumours has been established in the literature. In this study, even though on a relatively small patient cohort, we also found a significant positive prognostic role for stromal infiltration of cytotoxic T cells, Th1 cells and T regulatory cells at the tumour front of stage I–IV CRC patients, with the strongest association found for cytotoxic T cells (P = 0.009) (Supplementary Fig. S1). The prognostic significance remained in stage I-III CRC patients for cytotoxic T cells and T regulatory cells (P = 0.047 and P = 0.014, respectively). Stratifying the prognostic role of stromal infiltration of cytotoxic T cells by KRAS or BRAF mutation status in both the study cohort and the NSHDS validation cohort, revealed a prognostic role of cytotoxic T cells independent of KRAS and BRAF mutation in CRC (Fig. 2). The independent prognostic role of cytotoxic T cells was further tested in a multivariable Cox Proportional-Hazards model in stage I-III patients from the NSHDS cohort. The positive prognostic role of cytotoxic T cells remained statistically significant in a multivariable analysis including gender, age, localisation, tumour stage, BRAF mutation, KRAS mutation, and MSI status (Supplementary Table S5).

Fig. 2: The prognostic importance of cytotoxic T cells at the tumour front in CRC in relation to KRAS/BRAF mutation status.
figure 2

Shown are Kaplan–Meier plots of cancer-specific survival in patients with different levels of stromal infiltration of cytotoxic T cells (Tcyt) in wild-type/wild-type tumours, KRAS mutant tumours, and BRAF mutant tumours, as indicated, in (a) the study cohort, and in (b) the NSHDS validation cohort. Pre-operatively irradiated rectal cancers were excluded. Log-rank tests were used to calculate differences in 5-year survival between groups. P < 0.05 was considered statistically significant.

Full size image

Discussion

The immune response has important clinical value in both prognosis and therapy of CRC patients. Immune checkpoint inhibitors have proven very effective in some patients with metastatic immunogenic MSI tumours, but there are likely additional CRC patient groups where immunotherapy may have clinical value. CRC is a heterogeneous disease, which calls for careful investigations of the immune response in molecular subgroups of CRC to identify potential predictive markers of response to immunotherapy. Here, we investigated the infiltration of different immune cell subsets in relation to molecular characteristics of CRC, with a focus on the clinically well-established molecular markers KRAS and BRAF mutation.

We analysed the tumour infiltration of five different immune cells (cytotoxic T cells, Th1 cells, regulatory T cells, B cells and macrophages), both in the tumour centre and at the tumour front of patients with CRC. Immune cells were quite evenly distributed between the tumour centre and tumour front, however the associations to clinicopathological and tumour molecular traits were found to be stronger at the tumour front. We further investigated both the stromal area and the tumour epithelial area, noticing a higher density of immune cells in the stromal compartment, a finding in line with a recent study by Mezheyeuski et al., where they used a similar approach to characterise the immune landscape in CRC [26]. These results support the analysis of immune cell infiltration in TMAs taken at the tumour front as a suitable approach in CRC.

Right-sided colon cancers were more highly infiltrated by different subsets of T cells at the tumour front, a higher immune cell infiltration being a recognised feature of these tumours [27]. We further found a decreasing infiltration of all immune cells analysed from stage I to stage IV tumours, findings that may be related to tumour immune escape and metastasis. Pre-operatively radiated rectal tumours displayed a significantly lower infiltration of all immune cells analysed, which is in line with the reported radiosensitivity of lymphocytic cells [28]. Even though being an interesting patient group, they should likely be analysed separately as pre-operative radiotherapy may alter the anti-tumour immune response, introducing e.g., damage associated molecular patterns and rediscovery of tumour antigens [29]. Pre-operative radiotherapy has so far shown contradictory results in CRC, with both enhanced and reduced immune cell infiltration, calling for further investigation [30,31,32].

When analysing immune cell infiltration in association with known molecular parameters of CRC tumours, we found the expected and well-established associations of immune cells to MSI tumours [4, 10]. Immunogenomic studies have, however, shown that not only hypermutated tumours (MSI or POLE mutated) but also some non-hypermutated MSS tumours display high neoantigen load and tumour immune cell infiltration [33,34,35]. Interestingly, we here found an opposing role of KRAS and BRAF mutation on the immune response in CRC. While KRAS-mutated tumours displayed decreased immune cell infiltration compared to KRAS wild-type tumours, the opposite was found for BRAF-mutated tumours, differences that were most obvious for the cytotoxic T cells and the Th1 cells. For cytotoxic T cells, the findings were validated in the NSHDS cohort. A contrasting role of KRAS and BRAF mutation on immune cell infiltration in CRC tumours has been previously suggested [36, 37]. Ling et al. found a similar contrasting infiltration of Th1 cells according to KRAS and BRAF mutation in CRC [37], and Lal et al. found that KRAS- and BRAF-mutated CRCs were found in clusters with low and high expression of immune response genes, respectively [36]. KRAS-mutated CRCs have in previous studies been associated with supressed immune pathways and decreased immune cell infiltration [38,39,40]. BRAF mutation has been linked to the immunogenic MSI tumours [27], but our findings partly suggest that BRAF and MSI may to some extent have individual effects on the immune response in CRC. Further larger studies are needed to establish the role of BRAF mutation on immune cell infiltration in CRC. In a study by Cen et al., where they compared the immune microenvironment in BRAF-mutated tumours to that of BRAF wild-type tumours, an increased immune cell infiltration was found along with an increased expression of immune checkpoints, e.g., PD1, PD-L1 and CTLA4 [19]. Bolzacchini et al. further identified a gene expression based “hot/inflamed” immunoprofile in a high fraction (52%) of patients with advanced BRAF-mutant CRC, with only a partial overlap with MSI [41]. Additionally, 42% of the MSI tumours showed a “cold” immunoprofile. Barras et al. also suggested different subtypes of BRAF-mutant CRCs based on gene expression, finding differences in immune regulating genes as well as patient survival, regardless of MSI, suggesting that the BRAF-mutated tumours are not uniform [42]. Further studies dissecting the immune response to BRAF-mutated tumours in MSI as well as MSS CRCs, including also analyses of immune checkpoint molecules, are needed to find potential predictive markers for immunotherapy. For patients with BRAF-mutated MSI CRCs, combined treatment with immune checkpoint inhibitors and targeted therapies are currently under clinical evaluation. In a clinical trial, Pembrolizumab as first line treatment of advanced MSI metastatic CRC showed good efficacy regardless of BRAF mutation [43]. However, a decreased benefit of Pembrolizumab treatment was found in KRAS- or NRAS-mutated metastatic MSI patients [43]. Furthermore, combination therapy with Nivolumab and Ipilimumab (CTLA4 inhibitor), showed evident beneficial effects, especially for patients with BRAF-mutated tumours [44]. Interestingly, even though limited by few patients, an improved clinical response to Pembrolizumab was found in patients with BRAF-mutated treatment-resistant metastatic MSS CRCs [45]. Importantly, our findings show that there are highly immune infiltrated MSS tumours (being the majority of CRCs), which should be further explored in terms of possible implications for immunotherapy. One limitation of this study was that immune analyses were restricted to patients who underwent primary surgery. The potential predictive value of immune markers needs to be further assessed also in the neo-adjuvant setting using larger patient cohorts. In fact, pathological responses have been shown for neo-adjuvant immunotherapy in both MSI and MSS early-stage cancers [46].

Mechanistic insights on the regulation of immunity according to KRAS and BRAF mutation are limited. In a study by Liao et al. KRAS mutation was found to direct the tumour recruitment of myeloid derived suppressor cells through the IRF2-CXCL3-CXCR2 axis leading to immune therapy resistance in CRC [39]. In a previous in vitro study comparing the cytokine profile of KRAS mutant, BRAF mutant, and BRAF/KRAS wild-type Caco2 CRC cells, we found an increased expression of a Th1 related cytokine (CXCL10), and decreased expression of Th2 related cytokines (CCL2 and TGFB1) in BRAF-mutated cells, while the opposite pattern was shown for KRAS-mutated cells [37]. The results were confirmed in tumour tissues and linked to infiltration of Th1 cells, suggesting that these mutations may have different and direct effects on the immune response to CRC.

In this study, we further analysed the prognostic role of immune cell infiltration in CRC tumours and found a significantly improved survival for patients with a high number of infiltrating T cell subsets (the strongest effect seen with cytotoxic T cells). The prognostic importance of cytotoxic T cells and regulatory T cells remained also in stage I-III patients.

KRAS mutation has been associated with a poor prognosis in CRC [47, 48], and so has BRAF mutation, irrespective of MSI status, with BRAF-mutated MSS tumours showing the worst prognosis [18]. Here, survival analyses of infiltrating cytotoxic T cells at the tumour front showed a strong prognostic role for cytotoxic T cells in CRC independent of KRAS- and BRAF-mutational status. Further studies are needed to understand the prognostic importance of immune cell infiltration in these molecular subgroups of CRC.

The strengths of this study include the quantitative evaluation of multiple immune cells in different tumour locations of patients with CRC, comprehensive immune evaluations on a protein-based level being rare and an important complement to the immunogenomic studies. We were also able to validate our main findings in an independent larger cohort of CRC patients. Weaknesses of this study include small sample sizes in the primary patient cohort and the possibility that the use of TMAs do not fully reflect the heterogenous distribution of tumour infiltrating immune cells within the whole tumour. Furthermore, the analyses in this study were restricted to BRAFV600E mutation and KRAS mutations in codon 12 and 13. Even though covering the majority of KRAS and BRAF mutations in CRC, further studies taking additional mutations into account are needed.

Collectively, our findings suggest that combined evaluation of MSI status, KRAS and BRAF mutation, and immune cell infiltration (cytotoxic T cells) may provide added value compared to each variable alone and might have potential to help guide decisions on immunotherapy in CRC. In future management of CRC patients, individualised therapy taking both the molecular characteristics of the tumour and the tumour immune response into consideration will be of importance. An increased understanding of the actions of the immune response in molecular subgroups of CRC may lead to significant advances in personalised medicine, including the identification of important prognostic and predictive tools, as well as new targets for therapy.