You are here

Article Text

Article menu
Colon
Intratumoral T cell infiltration, MHC class I and STAT1 as biomarkers of good prognosis in colorectal cancer
  1. Jonathan A D Simpson1,
  2. Ahmad Al-Attar2,
  3. Nicholas F S Watson1,
  4. John H Scholefield1,
  5. Mohammad Ilyas3,
  6. Lindy G Durrant2
  1. 1Department of Gastrointestinal Surgery, Queen's Medical Centre , University of Nottingham, UK
  2. 2Academic Department of Clinical Oncology, Institute of Immunology, Infections and Immunity, City Hospital, University of Nottingham, UK
  3. 3Department of Pathology, Queen's Medical Centre, University of Nottingham, UK
  1. Correspondence to Professor L G Durrant, Academic Department of Clinical Oncology, Institute of Immunology, Infections and Immunity, University of Nottingham, City Hospital Campus, Nottingham NG5 1PB, UK; lindy.durrant{at}nottingham.ac.uk

Abstract

Objective To evaluate immunosurveillance/editing in colorectal cancer.

Design Transformation stimulates the production of interferon γ (IFNγ) which signals via the IFNγ receptor (IFNGR1) on tumours. This results in stimulation of nuclear STAT1 (nSTAT1), inhibition of tumour growth and upregulation of major histocompatibility complex (MHC) while promoting T cell extravasation. In contrast, downregulation of MHC class I by allele loss results in loss of T cell recognition. A tissue microarray of 462 colorectal tumours with mean follow-up of 42 months (range 1–116) was stained by immunohistochemistry for markers which predict immunosurveillance/editing.

Results The presence of a high level of intratumoral T cells (ITTC) correlated with improved survival compared with a low level of ITTC, with a mean difference in survival of 16.3 months (p=0.006). There was a direct correlation between nSTAT1 expression and ITTC (p<0.001). Patients whose tumours had a high level of ITTC and nSTAT1 survived 20 months longer than those whose tumours had a low level of ITTC and no nSTAT1. A strong correlation was seen between ITTC and MHC class I expression (p=0.0002). A mean survival advantage of 26.1 months was seen in patients whose tumours had strong MHC I expression and high levels of ITTC over those who had weak MHC I and low levels of ITTC (log-rank test=12.023, p=0.034). Both MHC I and ITTC are independent predictors of good survival.

Conclusions ITTC, nSTAT1 and strong MHC class I expression on tumours identify patients with improved survival and an intact tumour immune system that may benefit from immunotherapy. Conversely, loss of these markers identifies patients whose tumours have escaped immunosurveillance and are unlikely to benefit from immunotherapy.

  • Colorectal cancer
  • intra-tumoral T-cells
  • nuclear STAT-1
  • MHC-1
  • Tissue microarray

http://dx.doi.org/10.1136/gut.2009.194472

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Significance of this study

    What is already known about this subject?

    • T cell infiltration in colorectal cancer is an independent predictor of poor prognosis.

    • Downregulation of major histocompatibility complex (MHC) class I but not complete loss is an independent predictor of poor prognosis in colorectal cancer.

    What are the new findings?

    • The majority of colorectal tumours express interferon γ receptor (IFNGR); however, only a small proportion show nuclear expression of STAT1 which suggests that most of these tumours are not signalling via this receptor.

    • nSTAT1 is an independent predictor of good prognosis.

    • Intratumoral T cells (ITTC) are an independent predictor of good survival.

    • There was a strong correlation between nSTAT1 and ITCC and between ITCC and MHC class I expression.

    • When all factors are combined with standard clinicopathological variables, ITTC, MHC class I, TNM stage and vascular invasion are all independent predictors of patient survival.

    Introduction

    Worldwide, it is estimated that over a million new cases of colorectal cancer (CRC) are diagnosed yearly, accounting for more than 9% of all new cancer cases.1 It is also the second most common cause of cancer-related deaths. Despite the recent advances in treating cancer, the 5-year survival rate from CRC remains at 50% and 10% for TNM stages III and IV, respectively. Currently, there are four treatment options for CRC: surgery, chemotherapy, radiotherapy and monoclonal antibody therapy. These different modalities can be used alone or in combination, depending on the stage of disease and fitness of the patient. Despite the historical success of these modalities, none of them takes into account the individual tumour biology or the immune response to the cancer. It has been shown that transformation induces stress-related molecules which are recognised by innate immune cells with the subsequent release of interferon γ (IFNγ).2 Biological responses to IFNγ are mediated mainly by the regulation of gene expression via the JAK-STAT1 pathway. The IFNγ receptor comprises two subunits, IFNGR1 and IFNGR2. IFNγ binding induces receptor oligomerisation and activation of the receptor-associated Janus kinases, Jak1 and Jak2. These in turn phosphorylate the intracellular domain of the receptor which serves as a docking site for STAT1.3 Latent STAT1 in the cell cytoplasm then undergoes dimerisation and translocates to the nucleus where it regulates gene expression by binding to γ-activated sequence elements in the promoter regions of IFNγ-regulated genes. Several studies have underpinned the importance of the IFNγ pathway in the inhibition of tumour growth. Mice lacking IFNγ have an increased incidence of both chemically-induced and spontaneous tumour formation.4 Similarly, IFNGR or STAT1 knockout mice have an increased incidence of chemically-induced tumours, and p53/STAT1 double knockout mice also have an increased risk of spontaneous tumours.5 6 Furthermore, in contrast to IFN-α knockout mice where the response is orchestrated from the bone marrow, IFNγ asserts its effects directly on the tumour cell.7 It can directly inhibit tumour growth by stimulating production of the cell cycle inhibitors p21 and p27,8–10 stimulate production of caspase 1 and upregulation of Fas resulting in apoptosis,11 12 and it also induces upregulation of MHC molecules and secretion of chemokines to enhance immune recognition of tumours.13

    Recent studies have shown that the immune system can play an important role in CRC development and progression. Patients with CRC are able to mount an antigen-specific T cell response even without prior immunotherapy.14 Published data indicate that the presence of tumour-infiltrating T cells in CRC tissue predicts an improved prognosis.15–20 However, for an effective response, T cells need to engage the MHC molecule present on tumour cells. It is hypothesised that immune editing of MHC-positive tumours results in the emergence of tumour clones that downregulate specific MHC alleles to evade T cell recognition. We have previously shown that downregulation of MHC is an independent predictor of poor survival in CRC.21

    In this study we use high-throughput tissue microarray (TMA) technology to study the expression of intratumoral T cells (ITTC), IFNGR1 and nSTAT1 in a large well-characterised cohort of patients with CRC. The expression of these markers was correlated with MHC class I expression, survival and standard clinicopathological criteria.

    Materials and methods

    In an attempt to overcome some of the reporting deficiencies inherent in prognostic tumour marker studies, the REporting recommendations for tumour MARKer prognostic studies (REMARK)22 were followed.

    Patient cohort

    The study population cohort comprised a consecutive series of 462 archived specimens of primary invasive CRC tissue obtained from patients undergoing elective surgical resection of a histologically proven primary CRC at Nottingham University Hospitals, Nottingham, UK.

    The samples were collected between January 1994 and December 2000 from the established institutional tumour bank and were identified from the hospital archives. No cases were excluded unless the relevant clinicopathological material/data were unavailable. The mean follow-up period was 42 months (range 1–116) to ensure a sufficient duration of follow-up to allow meaningful assessment of the prognostic value of the markers examined. Follow-up was calculated from the date of resection of the primary tumour, and all surviving cases were censored for data analysis in December 2003. Patient and tumour characteristics for the cohort have been previously published.21 All patients underwent surgical resection; those with lymph node positive disease routinely received adjuvant chemotherapy with 5-fluorouracil and folinic acid.

    Following resection, all tumours received in the histopathology laboratory were incised, fixed immediately in formaldehyde and processed through to embedding in paraffin wax, ensuring optimal tissue fixation and preservation for histological examination. Tissue microarrays were built as described previously.21

    Immunohistochemistry

    Tissue sections were placed in a hot oven at 60°C for 30 min, dewaxed in xylene and then rehydrated in three baths of 100%, 90% and 70% ethanol. Endogenous peroxidase activity was blocked using a 0.3% solution of hydrogen peroxide in Tris-buffered saline (TBS). Antigen retrieval was achieved using a rotary microwave oven; the slides were immersed in citrate buffer (pH 6.0) and placed in the centre of the oven for 10 min at 800 W then for another 10 min at 300 W. The slides were cooled down immediately for 10 min with tap water, then placed on a humidity chamber and covered with TBS. Normal swine serum diluted 1:20 in TBS was added to block non-specific adsorption of the antibodies to the tissue. Mouse monoclonal antibodies anti-CD3 antibody (SP7, NeoMarkers, California, USA, 1:150 dilution), anti-STAT1 antibody (ST1-3D4, Zymed Labs, California, USA, 1:50 dilution) and anti-IFNGR1 antibody (Tebu-bio, Peterborough, UK, 1:100 dilution) were incubated with the tissue sections for 1 h at room temperature. Universal streptavidin-biotin-peroxidase and DAB kits (Dako Glostrup, Denmark) were used to detect specific antibody binding according to the manufacturer's instructions. The slides were finally counterstained with haematoxylin (Gill's formula, Sigma Dorset, UK), then dehydrated and mounted. Negative controls were done by omitting the primary antibody.

    Scoring

    TMA slides were captured digitally and stored as high-resolution image files. All markers were assessed independently by two assessors (AS and AA) who were blind to the clinicopathological data of the patients.

    STAT1, IFNGR1 and MHC I staining were evaluated semiquantitatively. Owing to loss of cores as the TMA is section/stained, 410 cores were stained for STAT1 and MHC I, 410 for IFNGR1 and 355 cores for T cells. Tumours were considered to show positive expression of IFNGR1 and/or STAT1 when at least 10% of viable tumour cells within the tumour core displayed unequivocal staining for the antigen of interest. The intensity of IFNGR1/STAT1 expression was further allocated a semiquantitative score using a 4-point scale representing absent, weak, moderate or strong staining.

    T cell counts were calculated per area of tumour seen in each core using the following formula:Number of intra-tumoral T cells(ITTC)=total number of ITT cellsArea of tumour(mm2)

    The number of ITTC in the tumours examined ranged from 0 to 114/mm2. The cases were grouped around the mean density of cells into two groups: low (≤15) ITTC and high (>15) ITTC. T cells were also assessed within the stroma. However, as there were many more T cells within the stroma than within the tumour nests, this was done semiquantitatively. Cases were classified according to the density of CD3+ cell infiltration per stromal area as sparse, moderate or dense.

    Clinicopathological data

    Accurate and contemporaneous data regarding the date and cause of death of patients included in the study were provided prospectively on a weekly basis from the UK Office for National Statistics and were available for >99% of the cases arrayed. The validity of these data was further confirmed by independent review of the case notes of deceased individuals. In all cases, disease-specific survival was used as the primary end point. Additional clincopathological variables considered in the data analysis were patient gender, tumour site (colon versus rectal), histological tumour type, histological grade, pathologically-determined tumour stage according to the UICC, TNM classification of malignant tumours and the presence of extramural vascular invasion within the resected tumour specimen. Histological factors more recently identified as having potentially significant prognostic value in CRC such as the presence of perineural invasion and tumour border configuration were not recorded routinely and hence were not considered in the data analysis.

    Statistical analysis

    Statistical analysis of study data was performed using the SPSS Version 15.0 for Windows (SPSS Inc). Patients whose deaths resulted from non-CRC-related causes were censored at the time of death. Survival analysis was performed using Kaplan–Meier curves and the statistical significance of differences in disease-specific survival (DSS) between groups was estimated using the log-rank test. The χ2 test was used to determine the association between T cell infiltration (low vs high) and traditional clinical variables. The Cox proportional hazards model was used for multivariate analysis in order to determine the relative risk and independent significance of individual factors. In all cases, p values <0.05 were considered statistically significant.

    Results

    Patient characteristics

    The patient cohort (see table 1 in online supplement) consisted of 226 men (58%) and 196 women (42%). The median age at the time of surgery was 72 years, consistent with a median age at diagnosis of CRC of 70–74 years in the UK.1 Fifteen percent of tumours arrayed were TNM stage I, 38% were TNM stage II, 33% were TNM stage III and 12% were TNM stage IV, comparable with recently published national figures.1 The majority of tumours (85%) were adenocarcinomas and 77% were of a moderate histological grade. Evidence of extramural vascular invasion was documented in 76% of all tumours. Among all the clinicopathological parameters scored, only tumour stage and the presence or absence of vascular invasion had a strongly significant influence on survival. This is consistent with previously published data.21

    At the time of censoring for data analysis, 49% of patients had died from their disease, 13% had died from other causes and 37% were alive, which is comparable with a national average of a 5-year survival for CRC in the UK of approximately 45%.1 The median DSS was 58.2 months.

    IFNGR1 expression

    As IFNγ is a primary mediator of immunosurveillance, loss of the receptor (IFNGR1) may indicate an escape mechanism. Analysis of IFNGR1 expression was possible in 410 of the 462 tumours in the TMA (88.7%). Of these, 405 (98.8%) expressed the IFNGR1 molecule, which was seen as a predominantly cytoplasmic/membranous staining pattern (figure 1). In contrast, no staining was observed in either the tumour cell nuclei or surrounding stromal tissue; 164 (40.0%) showed no or weak staining (categorised as a low level of staining) and 246 (60.0%) showed moderate/strong staining (categorised as a high level of staining).

    Figure 1
    Figure 1

    (A–C) Cores from a tumour showing weak, moderate and strong staining for interferon γ receptor (IFNGR1). All ×200 original magnification (insets ×400 magnification).

    STAT1 expression

    IFNγ signalling is relayed to the cell nucleus via homodimerisation of the STAT1 protein. Analysis of STAT1 expression was possible in 405 of the 462 tumours in the TMA (88.7%). In 345/405 of tumours analysed (85.1%) STAT1 was expressed in the cell cytoplasm and in 72/405 cases (17.7%) it was found in the nucleus. A highly significant association was noted between the presence of both nuclear and cytoplasmic STAT1 within the same tumour samples, as nuclear STAT1 expression was restricted only to a subset of 72 cores (20.9%) of cytoplasmic STAT1-positive cases (p<0.001). Furthermore, a significant association was seen between the intensity of cytoplasmic and nuclear STAT1 expression within individual tumours. Representative examples of positive and negative staining for each antigen are shown in figure 2.

    Figure 2
    Figure 2

    (A–C) Cores from a tumour demonstrating cytoplasmic and nuclear staining for STAT1 (B, C) and STAT1-negative core (A). All ×200 original magnification (insets ×400 magnification).

    Correlation of IFNGR1 and STAT1 with clinicopathological criteria

    As the majority of tumours were found to express IFNGR1, in order to examine possible correlations between IFNGR1 and standard clinicopathological variables, tumours were categorised as showing either a low or a high level of IFNGR1 expression. A significant association was noted between IFNGR1 status and tumour grade only, with higher levels of IFNGR1 expression observed more frequently in tumours of low histological grade (n=259, 63.1% of well/moderately differentiated tumours) than in high-grade tumours (n=151; 36.8% of poorly differentiated tumours; p=0.002).

    Correlation of clinicopathological data with expression of nSTAT1 showed a strong association between nSTAT1 expression and histological grade (p=0.003). However, there were no significant correlations between cytoplasmic STAT1 expression and any clinicopathological variables.

    Correlations between IFNGR1, cytoplasmic STAT1, nuclear STAT1 and DSS were assessed using Kaplan–Meier plots and log-rank testing (table 1). No significant difference in survival was noted between patients with tumours displaying either low (mean DSS 65.2 months) or high (mean DSS 66.4 months) expression of IFNGR1. However, the presence of nSTAT1 was seen to be significantly associated with an improved survival with a mean DSS of 86.6 months (95% CI 76.4 to 96.3) in patients with nSTAT1-positive tumours compared with a mean DSS of 70.0 months (95% CI 64.4 to 75.1) in nSTAT1-negative cases (p=0.008; figure 3). While cytoplasmic expression of STAT1 showed a trend towards a longer DSS, this was not significant (figure 3).

    View this table:
    Table 1

    Kaplan–Meier survival analysis of correlations between expression of interferon γ receptor (IFNGR1; n=410), cytoplasmic STAT1 (n=402), nuclear STAT1 (n=402) and disease-specific survival (DSS)

    Figure 3
    Figure 3

    Kaplan–Meier plots for disease-specific survival for (A) cytoplasmic STAT1 and (B) nuclear STAT-1.

    In order to assess the relative influence of nSTAT1 expression in conjunction with vascular invasion and TNM stage, a multivariate analysis was performed using the Cox proportional hazards model (table 2). Of the conventional clinicopathological variables assessed, both increasing TNM stage (p<0.001) and vascular invasion status (p=0.001) were identified as having independent prognostic significance. In this model, positive nSTAT1 expression was also found to confer tumour protection that was independent of clinicopathology, with a hazard ratio for survival in patients with nuclear STAT1-positive tumours of 0.48 (95% CI 0.28 to 0.81), p=0.007.

    View this table:
    Table 2

    Multivariate Cox regression model for vascular invasion, TNM stage and nuclear STAT1 expression

    Intratumoral T cells (ITTC) and stromal T cells

    The number of ITTC in the tumour nest was carefully quantified in 355 tumours (table 3). There were 121 patients (34.1%) in the low ITTC group (≤15 cells/mm2 of which 53 had no T cell infiltration) and 234 (65.9%) in the high ITTC group (>15 cells/mm2). Examples of cores with low and high ITTC levels are shown in figure 4. The density of stromal T cells varied widely with 141 (39.7%) cases showing sparse infiltration, 175 (49.3%) with moderate infiltration and only 39 (10.9%) cases showing dense CD3+ infiltration. All tumours had some T cells in their stroma. Examples of cores with low and high stromal CD3 staining are shown in figure 4.

    View this table:
    Table 3

    Expression level of intratumoral T cells (ITTC) and stromal T cells with mean survival data, standard errors (SE) and 95% CI for patients in all groups

    Figure 4
    Figure 4

    CD3 immunohistochemical staining of formalin-fixed paraffin-embedded colorectal cancer cores using the ABC method. Total core areas are indicated in red; specific tumour area in blue. Only cells within the blue perimeter are counted as intratumoral T cells (ITTCs). All other cells are considered stomal. (A) Core with high ITTC level; (B) core with low ITTC level.

    Survival prediction and correlations with clinicopathological criteria

    The presence of high ITTC levels correlated with improved survival compared with low ITTC levels, with a mean survival difference of 16.4 months (table 3). This was statistically significant, as shown in the Kaplan–Meier graph (figure 5A) and the log-rank test (p=0.002). The prevalence of ITTC was found to correlate with lower TNM stage (p=0.003), colonic tumours (p=0.044) and the absence of extramural vascular invasion (p=0.041; table 4). A multivariate Cox regression model incorporating the influences of stage and extramural vascular invasion with ITTC, adjusting for their collective effect, shows that the latter factor retained its significant influence on prognosis (HR=0.63, 95% CI 0.43 to 0.93), p=0.019 (table 5).

    View this table:
    Table 4

    Correlations of intratumoral T cell (ITTC) levels with clinicopathological variables, microsatellite status and major histocompatibility complex (MHC) expression

    View this table:
    Table 5

    Multivariate Cox regression model for vascular invasion, TNM stage and intratumoral T cell (ITTC) level

    Figure 5
    Figure 5

    Kaplan–Meier graph showing difference in survival probability between patients whose tumours have (A) low (≤15 cell/mm2) and high (>15 cell/mm2) intratumoral T cell levels (p=0.006); (B) sparse, moderate or dense stromal T cell infiltration (p=0.167).

    A strong positive correlation was found between stromal T cells and ITTCs (p<0.001), indicating that the more T cells found in the stroma, the more likely T cells were to be found in the tumour tissue. However, the prevalence of stromal T cells did not correlate with any of the clinicopathological variables. Stromal T cell density also correlated with survival (figure 5B), showing better survival with denser infiltrate (p=0.032), although this was not as significant as the ITTC.

    Correlation of ITTC with nSTAT1

    There was also a positive correlation between nSTAT1 expression and ITTC, with 62 tumours (86.1%) with positive nSTAT1 having a higher level of inflammatory cell infiltrate than those with no nSTAT1 (n=202; 61.2%; p<0.001). Patients whose tumours had high ITTC levels and nSTAT1 survived 19.7 months longer than tumours with low ITTC levels and no nSTAT1 (table 6). MHC expression, as measured by expression of both heavy and light chains, has been previously published.21 The majority of tumours expressing nSTAT1 also had strong levels of expression of MHC class I (45; 62%) compared with nSTAT1-negative tumours (148; 44.9%; p=0.029).

    View this table:
    Table 6

    Influence of intratumoral T cell (ITTC) level and nuclear STAT1 expression on mean disease-specific survival

    Correlation of ITTC with MHC class I

    A strong correlation was seen between ITTC and MHC I expression (p=0.0002; table 4). Twenty-three tumours (22%) with complete loss of MHC I had a low ITTC infiltrate while only 10 (10%) had a high level of ITTC. In contrast, 108 tumours (53.1%) that expressed the MHC I molecule showed a dense infiltrate by T cells. Survival times differed significantly between patient groups depending on the expression levels of MHC I and ITTC. A mean survival advantage of 26.1 months was seen in patients whose tumours had strong MHC I expression and a high ITTC infiltrate over those who had weak MHC I expression and low CD3 cell infiltrate. A breakdown of all groups is shown in table 7. The difference in survival between the groups was statistically significant (log-rank test=12.023, p=0.034).

    View this table:
    Table 7

    Mean disease-specific survival, standard error (SE) and 95% CI for patients with colorectal cancer in relation to major histocompatibility complex (MHC) class I and intratumoral T cell (ITTC) expression

    Correlation of nSTAT1, ITTC and MHC class I with prognosis

    As ITTC, loss of MHC class I and nSTAT1 where all shown to be predictors of a poor prognosis independent of classical clinicopathological characteristics, a multivariate analysis was performed to see if they were independent of each other. Both ITTC (p=0.009) and weak MHC I expression (p=0.038) retained prognostic significance but nSTAT1 was not independent of these factors (table 8). When ITTC (p=0.016) and MHC I (p=0.012) were combined with tumour stage (p=0.001) and vascular invasion (p=0.005), they both still retained their independent prognostic significance (table 9).

    View this table:
    Table 8

    Multivariate Cox regression analysis of major histocompatibility complex (MHC) class I, nuclear STAT1 and intratumoral T cell (ITTC) level

    View this table:
    Table 9

    Multivariate Cox regression model for vascular invasion, TNM stage, intratumoral T cell (ITTC) level, nSTAT1 and major histocompatibility complex (MHC) class I

    Discussion

    Tumour transformation upregulates stress molecules such as the NKG2D ligands. We have shown that one of these ligands, MICA, is an independent predictor of good prognosis in CRC.22 23 NKG2D ligands interact with the NKG2D receptor on NK cells or are recognised by the T cell receptor on a subset of NKT cells and stimulate the production of IFNγ. This cytokine interacts with its specific receptor IFNGR1 on cells stimulating intracellular signalling and nuclear localisation of STAT1. STAT1 regulates direct antitumour and antiproliferative/proapoptotic effects and an indirect effect by promoting cellular infiltration due to release of chemokines. This appears to be one of the mechanisms underlying T cell infiltration within the CRC tumour nests, as we have shown that increased ITTC correlated with a significantly higher expression of nSTAT1. In patients with tumours with a dense T cell infiltrate that express positive nSTAT1, a longer DSS was recorded. This suggests that IFNγ signalling plays a role in immunosurveillance of CRC.24 Lack of nSTAT1 in the majority of tumours could be due to lack of IFNγ signalling. However, defective IFNγ signalling may provide a selective advantage to tumours. This does not appear to be mediated by loss of the IFNGR1 receptor as all tumours expressed this receptor and low expression did not have any impact on survival. Lack of signalling may be mediated by functional STAT1 loss.25 Several cell lines have been shown to be deficient in STAT1 signalling due to either upregulation of STAT1 dephosphorylation and subsequent ubiquitin-mediated targeting of the transcription factor for proteosomal degradation, or dysregulation of STAT1 phosphorylation by the JAK kinases.6 11 Latent STAT1 in the cell cytoplasm undergoes dimerisation, then translocates to the nucleus where it regulates gene expression by binding to γ-activated sequence elements in the promoter regions of IFNγ-regulated genes. HLA heavy chain genes are downstream genes regulated by IFNγ. We have previously shown that partial loss of HLA expression is associated with a poor prognosis in CRC.26 In this study we show that this may be related to lack of STAT1 signalling as there was a positive correlation between nSTAT1 and MHC I expression. Loss of STAT1 signalling has also been shown to be associated with a higher incidence of tumours, implicating it as an intrinsic tumour suppressor molecule in mice.5 27 STAT1 is a key regulator of the response not only to type II but also to type I interferons, and is stimulated by growth factors and in response to DNA damage.28 Tumours which acquire resistance to IFNγ by dysregulation of STAT1 may therefore also be resistant to multiple cytokine/growth factors and to DNA repair mechanisms, contributing further to genetic instability and tumour aggression.

    For almost 20 years the degree of lymphocytic infiltrate in CRC has been proposed as a system for predicting prognosis.29 However, the results have been conflicting with groups staining leucocytes by H&E17 30 and by immunohistochemistry18 showing an association with improved survival, and other similar studies failing to show a positive correlation between CD3 cells and overall survival. However, recent studies on T cell localisation within tumours have shown differences between stroma and tumour nest-associated T cells and survival.31 In this study we show that the presence of high intratumoral T cell infiltration correlates with better survival and can predict the behaviour of CRC. There was a difference of 16.4 months in mean survival between patients with high and low intratumour infiltration. This finding confirms the role played by T lymphocytes in controlling CRC growth. Galon et al reported that, in the series of tumours they studied, the type, density and location of immune cells in CRC had a better prognostic value than TNM stage classification.31 Using immunofluorescence methods, Milasiene et al studied CD3+ cells in the blood of 73 patients diagnosed with primary colorectal and gastric cancer. They showed that high levels of CD3+ cells in patients with stage III CRC had better survival rates.32 Our results show that T cells located in the stroma were not as strong predictors of prognosis as T cells located within tumour nests. The inability of stromal T cells to make an impact on the growth of tumour cells is probably because T cells cannot mediate an effect without being in direct contact with cancer cells. T cells recognise tumour-associated peptides in association with MHC and hence require direct contact between target and effector cells. Menon et al proposed that there was a basement membrane-like structure surrounding tumour nests in 53% of patients with CRC which prevents contact between immune effector cells and target cancer cells.15 33 There was no association between microinstability (MSI) and T cell infiltration in this study. Although MSI may drive enhanced immune recognition, it may also drive immune editing leading to no improvement in T cell infiltration over more stable MSI-negative tumours.

    We found that the number of CD3+ cells within a tumour correlated inversely and strongly with MHC I expression levels (p=0.0002). High levels of MHC I expression make tumour cells good targets for T cells whereas weak expression (due to allele loss) prevents specific T cell recognition. In contrast, complete loss of MHC makes these cells good targets for NK lysis. Using the same cohort of patients, we now show that, if a tumour exhibits a high T cell infiltrate and strong expression of MHC I, the prognosis for the patient is better than for those who have a high T cell infiltrate but weak MHC I levels. Although nSTAT1, ITTC and MHC I are markers of patient prognosis, when combined in a multivariate analysis nSTAT1 loses its prognostic significance. In contrast, if ITTC and MHC I are combined with tumour stage and vascular invasion, they all retain their independent prognostic significance.

    In conclusion, this study shows that high levels of ITTC, nSTAT1 and high MHC I expression are associated with improved disease-specific survival and can identify a group of patients (66%) who may benefit from immunotherapeutic strategies such as cytokine or vaccine therapy. Low levels of ITTC, no nSTAT1 and weak MHC class I expression are associated with worse disease-specific survival, suggesting that these tumours are more aggressive as they have evaded the immune system and would not benefit from immune intervention. Thus, ITTC, nSTAT1 and MHC class I are useful biomarkers that have a role in the pathogenesis of cancer. Both ITTC and MHC I have been validated in other data sets as predictors of patient survival but, in order to really validate their use as biomarkers they need to be screened in randomised trials.34

    References

    1. ↵
      1. Quinn M,
      2. Babb P,
      3. Brock A,
      4. et al
      . Cancer trends in England and Wales 1950–1999. London: The Stationery Office, 2001.
    2. ↵
      1. Ramana CV,
      2. Gil MP,
      3. Schreiber RD,
      4. et al
      . Stat1-dependent and -independent pathways in IFN-gamma-dependent signalling. Trends Immunol 2002;23:96–101.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Chen X,
      2. Vinkemeier U,
      3. Zhao Y,
      4. et al
      . Crystal structure of a tyrosine phosphorylated STAT-1 dimer bound to DNA. Cell 1998;93:827–39.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Street SE,
      2. Trapani JA,
      3. MacGregor D,
      4. et al
      . Suppression of lymphoma and epithelial malignancies effected by interferon gamma. J Exp Med 2002;196:129–34.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Kaplan DH,
      2. Shankaran V,
      3. Dighe AS,
      4. et al
      . Demonstration of an interferon gamma-dependent tumor surveillance system in immunocompetent mice. Proc Natl Acad Sci USA 1998;95:7556–61.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Shankaran V,
      2. Ikeda H,
      3. Bruce AT,
      4. et al
      . IFNgamma and lymphocytes prevent primary tumour development and shape tumour immunogenicity. Nature 2001;410:1107–11.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Dunn GP,
      2. Bruce AT,
      3. Sheehan KC,
      4. et al
      . A critical function for type I interferons in cancer immunoediting. Nat Immunol 2005;6:722–9.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Bromberg JF,
      2. Horvath CM,
      3. Wen Z,
      4. et al
      . Transcriptionally active Stat1 is required for the antiproliferative effects of both interferon alpha and interferon gamma. Proc Natl Acad Sci USA 1996;93:7673–8.
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Chin YE,
      2. Kitagawa M,
      3. Su WC,
      4. et al
      . Cell growth arrest and induction of cyclin-dependent kinase inhibitor p21 WAF1/CIP1 mediated by STAT1. Science 1996;272:719–22.
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Mandal M,
      2. Bandyopadhyay D,
      3. Goepfert TM,
      4. et al
      . Interferon-induces expression of cyclin-dependent kinase-inhibitors p21WAF1 and p27Kip1 that prevent activation of cyclin-dependent kinase by CDK-activating kinase (CAK). Oncogene 1998;16:217–25.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Chin YE,
      2. Kitagawa M,
      3. Kuida K,
      4. et al
      . Activation of the STAT signaling pathway can cause expression of caspase 1 and apoptosis. Mol Cell Biol 1997;17:5328–37.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Xu X,
      2. Fu XY,
      3. Plate J,
      4. et al
      . IFN-gamma induces cell growth inhibition by Fas-mediated apoptosis: requirement of STAT1 protein for up-regulation of Fas and FasL expression. Cancer Res 1998;58:2832–7.
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Sgadari C,
      2. Farber JM,
      3. Angiolillo AL,
      4. et al
      . MIG, the monokine induced by interferon-gamma, promotes tumor necrosis in vivo. Blood 1997;89:2635–43.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Nagorsen D,
      2. Scheibenbogen C,
      3. Marincola FM,
      4. et al
      . Natural T cell immunity against cancer. Clin Cancer Res 2003;9:4296–303.
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Menon AG,
      2. Janssen-van Rhijn CM,
      3. Morreau H,
      4. et al
      . Immune system and prognosis in colorectal cancer: a detailed immunohistochemical analysis. Lab Invest 2004;84:493–501.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Morris M,
      2. Platell C,
      3. Iacopetta B
      . Tumor-infiltrating lymphocytes and perforation in colon cancer predict positive response to 5-fluorouracil chemotherapy. Clin Cancer Res 2008;14:1413–7.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Ropponen KM,
      2. Eskelinen MJ,
      3. Lipponen PK,
      4. et al
      . Prognostic value of tumour-infiltrating lymphocytes (TILs) in colorectal cancer. J Pathol 1997;182:318–24.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Naito Y,
      2. Saito K,
      3. Shiiba K,
      4. et al
      . CD8+ T cells infiltrated within cancer cell nests as a prognostic factor in human colorectal cancer. Cancer Res 1998;58:3491–4.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Diederichsen AC,
      2. Hjelmborg JB,
      3. Christensen PB,
      4. et al
      . Prognostic value of the CD4+/CD8+ ratio of tumour infiltrating lymphocytes in colorectal cancer and HLA-DR expression on tumour cells. Cancer Immunol Immunother 2003;52:423–8.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Funada Y,
      2. Noguchi T,
      3. Kikuchi R,
      4. et al
      . Prognostic significance of CD8+ T cell and macrophage peritumoral infiltration in colorectal cancer. Oncol Rep 2003;10:309–13.
      OpenUrlPubMedWeb of Science
    21. ↵
      1. Watson NF,
      2. Ramage JM,
      3. Madjd Z,
      4. et al
      . Immunosurveillance is active in colorectal cancer as downregulation but not complete loss of MHC class I expression correlates with a poor prognosis. Int J Cancer 2006;118:6–10.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Watson NF,
      2. Spendlove I,
      3. Madjd Z,
      4. et al
      . Expression of the stress-related MHC class I chain-related protein MICA is an indicator of good prognosis in colorectal cancer patients. Int J Cancer 2006;118:1445–52.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. McGilvray RW,
      2. Eagle RA,
      3. Watson NF,
      4. et al
      . NKG2D ligand expression in human colorectal cancer reveals associations with prognosis and evidence for immunoediting. Clin Cancer Res 2009;15:6993–7002.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. McShane LM,
      2. Altman DG,
      3. Sauerbrei W,
      4. et al
      . REporting recommendations for tumour MARKer prognostic studies (REMARK). Br J Cancer 2005;93:387–91.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Dunn GP,
      2. Old L,
      3. Schreiber RD
      . The three es of cancer immunoediting. Annu Rev Immunol 2004;22:329.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Chen B,
      2. He L,
      3. Savell VH,
      4. et al
      . Inhibition of the interferon-gamma/signal transducers and activators of transcription (STAT) pathway by hypermethylation at a STAT-binding site in the p21WAF1 promoter region. Cancer Res 2000;60:3290–8.
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. Levy DE,
      2. Gilliland DG
      . Divergent roles of STAT1 and STAT5 in malignancy as revealed by gene disruptions in mice. Oncogene 2000;19:2505–10.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Townsend PA,
      2. Cragg MS,
      3. Davidson SM,
      4. et al
      . STAT-1 facilitates the ATM activated checkpoint pathway following DNA damage. J Cell Sci 2005;118:1629–39.
      OpenUrlAbstract/FREE Full Text
    29. ↵
      1. Jass JR,
      2. Love SB,
      3. Northover JM
      . A new prognostic classification of rectal cancer. Lancet 1987;1:1303–6.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Di Giorgio A,
      2. Botti C,
      3. Tocchi A,
      4. et al
      . The influence of tumor lymphocytic infiltration on long term survival of surgically treated colorectal cancer patients. Int Surg 1992;77:256–60.
      OpenUrlPubMedWeb of Science
    31. ↵
      1. Galon J,
      2. Fridman WH,
      3. Pages F
      . The adaptive immunologic microenvironment in colorectal cancer: a novel perspective. Cancer Res 2007;67:1883–6.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Milasiene V,
      2. Stratilatovas E,
      3. Norkiene V
      . The importance of T-lymphocyte subsets on overall survival of colorectal and gastric cancer patients. Medicina (Kaunas, Lithuania) 2007;43:548–54.
      OpenUrlPubMed
    33. ↵
      1. Menon AG,
      2. Fleuren GJ,
      3. Alphenaar EA,
      4. et al
      . A basal membrane-like structure surrounding tumour nodules may prevent intraepithelial leucocyte infiltration in colorectal cancer. Cancer Immunol Immunother 2003;52:121–6.
      OpenUrlPubMed
    34. ↵
      1. Bucher HC,
      2. Guyatt GH,
      3. Cook DJ,
      4. et al
      . Users' guides to the medical literature: XIX. Applying clinical trial results. A. How to use an article measuring the effect of an intervention on surrogate end points. Evidence-Based Medicine Working Group. JAMA 1999;282:771–8.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    Footnotes

    • Funding The Lewis Trust, the Mason Medical Small Grants Fund and the Royal College of Surgeons of Edinburgh.

    • Competing interests None.

    • Ethics approval This study was conducted with the approval of the Queen's Medical Centre.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    Linked Articles

    • Digest
      Digest
      Emad El-Omar Severine Vermeire Alexander Gerbes
      Gut 2010; 59 i-ii Published Online First: 25 Jun 2010. doi: 10.1136/gut.2010.219337