You are here

Article Text

Article menu

Download PDFPDF

Colon
Original article
Tumour CD274 (PD-L1) expression and T cells in colorectal cancer
  1. Yohei Masugi1,
  2. Reiko Nishihara1,2,3,4,
  3. Juhong Yang1,5,
  4. Kosuke Mima1,
  5. Annacarolina da Silva1,
  6. Yan Shi1,
  7. Kentaro Inamura6,
  8. Yin Cao2,7,8,
  9. Mingyang Song2,7,8,
  10. Jonathan A Nowak9,
  11. Xiaoyun Liao1,10,
  12. Katsuhiko Nosho11,
  13. Andrew T Chan7,8,12,
  14. Marios Giannakis1,13,14,
  15. Adam J Bass1,13,14,
  16. F Stephen Hodi1,10,
  17. Gordon J Freeman1,14,
  18. Scott Rodig15,
  19. Charles S Fuchs1,12,
  20. Zhi Rong Qian1,
  21. Shuji Ogino1,3,9
  1. 1 Department of Medical Oncology, Dana-Farber Cancer Institute and Harvard Medical School, Boston, Massachusetts, USA
  2. 2 Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA
  3. 3 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA
  4. 4 Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA
  5. 5 Collaborative Innovation Center of Tianjin for Medical Epigenetics, Key Laboratory of Hormone and Development, Metabolic Disease Hospital & Tianjin Institute of Endocrinology, Tianjin Medical University, Tianjin, China
  6. 6 Division of Pathology, Cancer Institute, Japanese Foundation for Cancer Research, Tokyo, Japan
  7. 7 Clinical and Translational Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts, USA
  8. 8 Division of Gastroenterology, Massachusetts General Hospital, Boston, Massachusetts, USA
  9. 9 Division of MPE Molecular Pathological Epidemiology, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
  10. 10 Center for Immuno-Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts, USA
  11. 11 Department of Gastroenterology, Rheumatology and Clinical Immunology, Sapporo Medical University School of Medicine, Sapporo, Japan
  12. 12 Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
  13. 13 Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA
  14. 14 Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
  15. 15 Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
  1. Correspondence to Dr Shuji Ogino, Division of MPE Molecular Pathological Epidemiology, Department of Pathology, Brigham and Women's Hospital, 450 Brookline Avenue, DFCI Room M422, Boston, MA 02215 USA; shuji_ogino{at}dfci.harvard.edu

Abstract

Objective Evidence suggests that CD274 (programmed death-ligand 1, B7-H1) immune checkpoint ligand repress antitumour immunity through its interaction with the PDCD1 (programmed cell death 1, PD-1) receptor of T lymphocytes in various tumours. We hypothesised that tumour CD274 expression levels might be inversely associated with T-cell densities in colorectal carcinoma tissue.

Design We evaluated tumour CD274 expression by immunohistochemistry in 823 rectal and colon cancer cases within the Nurses' Health Study and Health Professionals Follow-up Study. We conducted multivariable ordinal logistic regression analyses to examine the association of tumour CD274 expression with CD3+, CD8+, CD45RO (PTPRC)+ or FOXP3+ cell density in tumour tissue, controlling for potential confounders including tumour status of microsatellite instability (MSI), CpG island methylator phenotype, long interspersed nucleotide element-1 methylation level and KRAS, BRAF and PIK3CA mutations.

Results CD274 expression in tumour cells or stromal cells (including immune cells) was detected in 731 (89%) or 44 (5%) cases, respectively. Tumour CD274 expression level correlated inversely with FOXP3+ cell density in colorectal cancer tissue (outcome) (ptrend=0.0002). For a unit increase in outcome quartile categories, multivariable OR in the highest (vs lowest) CD274 expression score was 0.22 (95% CI 0.10 to 0.47). Tumour CD274 expression was inversely associated with MSI-high status (p=0.001). CD274 expression was not significantly associated with CD3+, CD8+ or CD45RO+ cell density, pathological lymphocytic reactions or patient survival prognosis.

Conclusions Tumour CD274 expression is inversely associated with FOXP3+ cell density in colorectal cancer tissue, suggesting a possible influence of CD274-expressing carcinoma cells on regulatory T cells in the tumour microenvironment.

  • MOLECULAR PATHOLOGY
  • CANCER EPIDEMIOLOGY
  • COLORECTAL CANCER
  • IMMUNE RESPONSE
  • T LYMPHOCYTES

https://doi.org/10.1136/gutjnl-2016-311421

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 on this subject?

    • The immune checkpoint PDCD1 (programmed cell death 1) pathway-targeted immunotherapy has emerged as a promising therapeutic strategy in various tumour types.

    • Tumour CD274 (programmed death-ligand 1) expression appears to suppress T-cell-mediated immune response against tumour through its interaction with T-cell co-receptor PDCD1.

    • High densities of various types of T cells in colorectal carcinoma tissue are associated with better clinical outcome.

    What are the new findings?

    • Tumour CD274 expression level is inversely associated with the density of FOXP3+ lymphocytes in colorectal cancer tissue, independent of potential confounders including tumour molecular status of microsatellite instability (MSI), CpG island methylator phenotype, long interspersed nucleotide element-1 methylation level and KRAS, BRAF and PIK3CA mutations.

    • Tumour CD274 expression was inversely associated with high-level MSI status in colorectal cancer.

    • Tumour CD274 expression level is not significantly associated with the density of CD3+, CD8+ or CD45RO+ cells in colorectal cancer tissue.

    How might it impact on clinical practice in the foreseeable future?

    • Our human population-based data suggest a possible influence of tumour CD274 expression on regulatory T cells in the tumour immune microenvironment. Hence, our current study can likely inform translational research on the development of immunotherapy strategies against colorectal cancer.

    Introduction

    Immunotherapy has emerged as a promising strategy to treat various types of cancers.1 Accumulating evidence indicates that the immune checkpoint mechanism play an important role in suppressing antitumour T-cell-mediated immune response in the tumour microenvironment.1 Studies have shown that therapeutic antibodies targeting the PDCD1 (programmed cell death 1, PD-1) protein and the CD274 (PDCD1 ligand 1, programmed death-ligand 1 (PD-L1), B7-H1) protein are effective in a number of cancer types.1–3 Emerging evidence also suggests complex roles of tumour molecular alterations and tumour–host interactions that influence response to these T-cell-based immunotherapies.4–6

    Colorectal carcinogenesis is driven by sequential genetic and epigenetic alterations of tumour cells and is also influenced by tumour–host interactions.7–14 A strong histological lymphocytic reaction, high density of CD3+ T cells and high densities of T-cell subpopulations (CD8+ cells, CD45RO (PTPRC)+ cells and FOXP3+ cells) in colorectal cancer tissue have been generally associated with favourable clinical outcome, supporting a major role of T-cell-mediated immunity in repressing tumour progression.14–19 Studies have shown that the abundance of tumour-infiltrating T cells is associated with specific molecular features of colorectal carcinoma, including high-level microsatellite instability (MSI-high).4 ,13–15 ,20 However, little is known on the complex interrelationship among tumour CD274 expression, tumour-infiltrating T cells and major tumour molecular features. In the colorectal cancer microenvironment, CD274-expressing tumour cells may inhibit antitumour activity of T cells. Hence, we hypothesised that tumour CD274 expression levels might be inversely associated with T-cell densities in colorectal cancer tissue. Because any in vitro or non-human system cannot perfectly recapitulate the complexity of human tumour or immune system, analyses of tumour characteristics and immune cells in human cancer tissue are valuable.

    To test our hypothesis, we analysed the two US nationwide prospective cohort studies, and examined tumour CD274 expression in relation to histological lymphocytic reaction or to CD3+, CD8+, CD45RO+ or FOXP3+ cell density in colorectal cancer tissue. Our comprehensive database including tumour immunity status and relevant clinicopathological and tumour molecular characteristics enabled us to investigate the independent association of tumour CD274 expression with T cells in tumour tissue, controlling for potential confounders.

    Materials and methods

    Study group

    We used two independent prospective cohort studies: the Nurses' Health Study (involving 121 701 women followed since 1976) and the Health Professionals Follow-up Study (involving 51 529 men followed since 1986).21–23 Every 2 years, the participants have been sent follow-up questionnaires to update information on potential disease risk factors, and to identify newly diagnosed cancers and other diseases in themselves and their first-degree relatives. The National Death Index was used to ascertain deaths of study participants and identify unreported lethal colorectal cancer cases. Study physicians reviewed all medical records related to colorectal cancer, extracted clinical information including the American Joint Committee on Cancer tumour, node, metastases (TNM) classification, the numbers of positive and negative lymph nodes harvested and tumour location, and determined cause of death in deceased individuals. In survival analyses, patients were followed until death or January 2012, whichever came first. We collected formalin-fixed paraffin-embedded (FFPE) tissue blocks from hospitals across the USA, where participants with colorectal cancer had undergone tumour resection. We included both colon and rectal carcinoma cases, considering the colorectal continuum model.24 We excluded polyposis syndrome cases, IBD-related cancers and cases that had received preoperative therapy. The study pathologist (SO) blinded to other data performed centralised pathology review, and recorded features including tumour differentiation, the extents of extracellular mucin, signet ring cells and solid tumour areas, tumour growth pattern and four patterns of histological lymphocytic reaction (Crohn's-like lymphoid reaction, peritumoural lymphocytic reaction, intratumoural periglandular reaction and tumour-infiltrating lymphocytes (TIL)).20 Tumour differentiation was categorised as well to moderate (>50% glandular area) or poor (≤50% glandular area). On the basis of the availability of CD274 expression data, a total of 823 colorectal cancer cases diagnosed up to 2008 were included in this study. In addition, 10 cases of colorectal carcinoma were selected from the archival pathology file of the Brigham and Women's Hospital, and anonymised, to perform immunohistochemistry (IHC) for CD274 in whole tissue sections. Tissue collection and analyses were approved by the human subjects committee at the Harvard T.H. Chan School of Public Health and the Brigham and Women's Hospital.

    Immunohistochemistry

    We constructed tissue microarray (TMA) from colorectal cancer blocks,25 and conducted IHC. Immunohistochemical analyses were performed for CD3, CD8, CD45RO and FOXP3, as previously described.15 We used an automated scanning microscope and the Ariol image analysis system (Genetix, San Jose, California, USA) to measure CD3+, CD8+, CD45RO+ and FOXP3+ cell densities in tumour tissue. We evaluated up to four TMA cores from each tumour, and calculated the average density (cells/mm2) of each T-cell population.15 Immunohistochemical method for PDCD1 (PD-1) is described in online supplementary material.

    For CD274 (PD-L1) IHC, tissue sections were deparaffinised, rehydrated and heated in a microwave for 15 min in Antigen Retrieval Citra Solution, pH 6 (BioGenex Laboratories, San Ramon, California, USA). Sections were incubated with Dual Endogenous Enzyme Block (Dako, Glostrup, Denmark), then were treated with Protein Block Serum-Free (Dako). Slides were incubated for 16 h at 4°C with a mouse monoclonal anti-CD274 antibody (Clone MIH1, eBioscience, San Diego, California, USA; dilution, 1:50). The primary antibody was visualised using EnVision+ System-HRP (Dako) for 30 min with diaminobenzidine, and counterstained with haematoxylin. Sections processed with replacement of primary antibody by Mouse IgG1K Isotype Control (Clone P3.6.2.8.1, eBioscience) were used as a negative control. Immunohistochemical expression for CD274 was interpreted by a pathologist (YM) unaware of other data. Tumour CD274 expression was evaluated based on immunostaining in the cytoplasm and membrane of tumour cells. Cytoplasmic expression level (intensity) was scored as 0 (absent), 1 (weak), 2 (moderate) or 3 (strong), and membrane expression level was scored as 0 (absent) or 1 (present; if distinct membrane staining above cytoplasmic staining level existed) (figure 1A–E). If staining difference was observed across multiple TMA tumour cores in each case, intensity of predominant staining pattern in tumour component was recorded. The overall tumour CD274 expression score was the sum of the cytoplasmic and membrane scores, ranging from 0 to 4. We scored CD274 expression of stromal cells including immune cells, based on the presence or absence of distinct cytoplasmic and/or membrane staining in stromal cells. Immune cells, including lymphocytes and macrophages, in the lymphoid tissue within normal colorectal tissue cores served as positive controls (figure 1F). A sample of 148 tumours was examined by a second pathologist (AdS). The agreement between the two observers for the tumour CD274 expression score was good with a weighted κ of 0.65 (95% CI 0.57 to 0.73). In addition, to assess intratumoural heterogeneity of CD274 expression, we performed CD274 IHC using whole tissue sections from the 10 selected colorectal carcinoma cases, and did not observe considerable intratumoural heterogeneity in any of the cases in terms of tumour CD274 expression (see online supplementary figure S1).

    Figure 1
    Figure 1

    Tumour CD274 expression in colorectal cancer. Tumour CD274 expression was evaluated based on immunostaining in the cytoplasm and membrane of tumour cells. Cytoplasmic expression level was scored as 0 (A), 1 (B), 2 (C) or 3 (D), according to cytoplasmic intensity. (E) Membrane CD274 expression in tumour cells. Membrane expression level was scored as 0 (absent) or 1 (present; if distinct membrane staining above cytoplasmic staining level existed). (F) Immune cells, including lymphocytes and macrophages, in the lymphoid tissue within normal colorectal tissue cores served as positive controls.

    Analyses of MSI, DNA methylation and KRAS, BRAF and PIK3CA mutations

    DNA was extracted from archival colorectal cancer tissue blocks. MSI status was analysed with the use of 10 microsatellite markers (BAT25, BAT26, BAT40, D2S123, D5S346, D17S250, D18S55, D18S56, D18S67 and D18S487), as previously described.22 ,26 We defined MSI-high as the presence of instability in ≥30% of the markers, and MSI-low/microsatellite stability (MSS) as instability in <30% of the markers.22 ,26 Methylation analyses of long interspersed nucleotide element-1 (LINE-1) and eight promoters specific for CpG island methylator phenotype (CIMP) (CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3 and SOCS1) were performed.22 ,27 CIMP-high, CIMP-low and CIMP-negative were defined as ≥6/8, 1/8–5/8 and 0/8 methylated promoters, respectively, according to the previously established criteria.22 ,27 PCR reaction and pyrosequencing were performed for BRAF (codon 600),26 KRAS (codons 12, 13, 61 and 146)28 and PIK3CA (exons 9 and 20).22 ,29

    Analysis of the amount of Fusobacterium nucleatum DNA

    We extracted DNA from colorectal carcinoma FFPE tissue sections, and performed a quantitative PCR assay to measure the amount of tissue Fusobacterium nucleatum DNA.30 ,31 We categorised colorectal carcinoma cases with detectable F. nucleatum DNA as low or high in relation to the median cut point amount of F. nucleatum DNA.30 ,31

    Statistical analysis

    All statistical analyses were conducted using SAS (V.9.3, SAS Institute, Cary, North Carolina, USA) and all p values were two-sided. Our primary hypothesis testing was an assessment of the association of the tumour CD274 expression score (an ordinal predictor variable) with each of (T) lymphocyte variables as outcome variables (CD3+, CD8+, CD45RO+ and FOXP3+ cell densities in colorectal cancer tissue, and the four histological lymphocytic reaction patterns). Because we examined the eight outcome variables, we adjusted two-sided α level to 0.006 (=0.05/8) by simple Bonferroni correction. All other analyses, including evaluation of individual OR estimates, represented secondary analyses. In those secondary analyses, in view of multiple comparisons, we interpreted our data cautiously, in addition to the use of the adjusted α level of 0.006.

    We performed multivariable logistic regression analysis to control for potential confounders. The multivariable model initially included age (continuous), sex, year of diagnosis (continuous), family history of colorectal cancer in a first-degree relative (present vs absent), tumour location (proximal colon vs distal colon vs rectum), MSI status (MSI-high vs MSI-low/MSS), CIMP status (high vs low/negative), BRAF mutation (mutant vs wild-type), KRAS mutation (mutant vs wild-type), PIK3CA mutation (mutant vs wild-type) and LINE-1 methylation level (continuous). A backward elimination with a threshold of p=0.05 was used to select variables in the final model. For cases with missing information in any of the categorical variables, we included those cases in the majority category of a given covariate. We assessed the proportional odds assumption in the ordinal logistic regression model, which was generally satisfied for the T-cell ordinal outcome variables (p>0.05), but not for three of the four patterns (peritumoural lymphocytic reaction, intratumoural periglandular reaction and TIL) of histological lymphocytic reaction ordinal variables (p≤0.032). Hence, we used the binary histological lymphocytic reaction variables as outcomes for logistic regression analysis.

    To assess the associations between tumour CD274 expression level and other categorical variables (except for the extents of signet ring cells and solid tumour areas, for which Fisher's exact test was performed), the χ2 test was performed. To compare mean age and mean LINE-1 methylation levels, an analysis of variance assuming equal variances was performed. All of those cross-sectional analyses for clinical, pathological, and molecular associations were secondary exploratory analyses, with adjusted two-sided α level of 0.002 (=0.05/22) for multiple hypothesis testing.

    Deaths from causes other than colorectal cancer were censored in colorectal cancer-specific mortality analyses. We performed Kaplan–Meier analysis, and log-rank test was used to compare survival between patient groups. To adjust for confounding, we used Cox proportional hazards regression models, and calculated HR for mortality. The multivariable models initially included disease stage (I/II vs III/IV/unknown) and the same set of variables as in multivariable logistic regression analysis. Variables were selected in a final multivariable model, using backward elimination with a threshold of p=0.05. To limit the degrees of freedom of the models, with missing information in any of the categorical covariates (tumour location (0.5%), MSI status (2.9%), CIMP status (8.3%), LINE-1 methylation level (2.8%), BRAF mutation (2.5%), KRAS mutation (2.9%) and PIK3CA mutation (8.7%)), we included those cases in the majority category of a given covariate. We confirmed that excluding the cases with missing information in any of the covariates did not substantially alter results (data not shown). We tested statistical interaction by the Wald test on the cross-product term of tumour CD274 expression score (ordinal categories ranging from 0 to 4) and each T-cell density variable (ordinal quartile categories) in a Cox proportional hazards regression model.

    Results

    We examined expression levels of the CD274 (PD-L1) protein in 823 cases of colorectal carcinoma within the two US nationwide prospective cohort studies. We scored tumour CD274 expression levels in cytoplasm (intensity, ranging from 0 to 3) and membrane (absent (0) or present (1) if distinct membrane staining above cytoplasmic staining level existed) (figure 1 and see online supplementary table S1). We used the sum of the cytoplasmic and membrane scores (ranging from 0 to 4) in each case for further analyses. Among the 823 colorectal cancer cases, 92 (11%), 234 (28%), 216 (26%), 238 (29%) and 43 (5%) tumours had tumour CD274 expression score of 0, 1, 2, 3 and 4, respectively.

    In 44 (5%) of the 823 colorectal cancer cases, CD274 expression was detectable in stromal cells including lymphocytes and macrophages in tumour stroma. There was no appreciable or significant association between the presence of CD274-expressing stromal cells and any of the clinical, pathological and tumour molecular features examined (see online supplementary table S2).

    Clinical, pathological and molecular characteristics according to the tumour CD274 expression score in colorectal cancer are summarised in table 1 (see online supplementary table S3 with 5 ordinal categories of tumour CD274 expression variable).

    View this table:
    Table 1

    Clinical, pathological and molecular features according to the tumour CD274 expression score in 823 colorectal cancer cases

    We observed inverse associations of tumour CD274 expression with tumour MSI status (p=0.001) and the extent of extracellular mucin (p<0.0001). The tumour CD274 expression level was not significantly associated with any of the other characteristics examined (p≥0.003; with the adjusted α level of 0.002 for multiple hypothesis testing) (table 1).

    Table 2 shows the distribution of colorectal carcinoma cases according to the tumour CD274 expression score, T-cell densities and histological lymphocytic reaction patterns. Tumour CD274 expression score was inversely correlated with FOXP3+ cell density (p<0.0001, by Spearman correlation test) and TIL (p=0.006, by Spearman correlation test) with the adjusted α level of 0.006.

    View this table:
    Table 2

    Distribution of colorectal cancer cases according to the tumour CD274 expression score, the densities of T cells and histological lymphocytic reaction patterns

    In our primary hypothesis testing, we conducted logistic regression analyses to assess the associations of the tumour CD274 expression score (an ordinal predictor variable) with CD3+, CD8+, CD45RO+ or FOXP3+ cell density (an ordinal quartile outcome variable) in colorectal cancer tissue (table 3, see online supplementary table S4). Tumour CD274 expression score was inversely associated with FOXP3+ cell density in both univariable and multivariable analyses (ptrend≤0.0002). For a unit increase in quartile categories of FOXP3+ cell density, the multivariable OR in the highest tumour CD274 expression score compared with the lowest score was 0.22 (95% CI 0.10 to 0.47). The tumour CD274 expression score was not significantly associated with CD3+, CD8+ or CD45RO+ cell density (all ptrend≥0.007, with adjusted α level of 0.006). Similar results were observed when we used binary T-cell density variables (see online supplementary tables S5 and S6).

    View this table:
    Table 3

    Ordinal logistic regression analysis to assess the association of the tumour CD274 expression score (predictor) with the density of T cells (outcome)

    We also examined the relationship of tumour CD274 expression levels with histological lymphocytic reaction (a binary outcome variable) (table 4). Although tumour CD274 expression was inversely associated with TIL in univariable analysis (ptrend=0.004), the association was not significant in multivariable analysis (ptrend=0.049) with the adjusted α level of 0.006. The tumour CD274 expression score was not significantly associated with any of the other histological lymphocytic reaction patterns.

    View this table:
    Table 4

    Logistic regression analysis to assess the association of the tumour CD274 expression score (predictor) with histological lymphocytic reaction (outcome)

    We assessed PDCD1 (PD-1) expression by IHC in 793 cases of colorectal carcinoma, and semiquantitatively scored PDCD1+ cell density in tumour tissue. Among the 793 cases, PDCD1+ cell density was scored as absent, very low, low, intermediate and high, in 230 (29%), 194 (24%), 140 (18%), 102 (13%) and 127 (16%) cases, respectively (see online supplementary figure S2). Clinicopathological and molecular characteristics according to PDCD1+ cell density in colorectal cancer are summarised in online supplementary table S7.

    As exploratory analyses to assess a prognostic role of CD274 expression in colorectal cancer, we conducted Kaplan–Meier analysis and Cox proportional hazards regression analysis, and did not observe significant association of tumour CD274 expression score with colorectal cancer-specific or overall mortality (table 5 and see online supplementary figure S3).

    View this table:
    Table 5

    Tumour CD274 expression and colorectal cancer patient mortality

    There was no significant association between CD274 expression in stromal cells and colorectal cancer mortality (see online supplementary table S8 and figure S4). We further examined whether prognostic associations of T-cell densities in tumour might be modified by tumour CD274 expression status, and did not observe significant interaction between tumour CD274 expression and T-cell (CD3+, CD8+, CD45RO+ or FOXP3+ cell) density (pinteraction>0.10) (see online supplementary table S9).

    Discussion

    We conducted this study to test the hypothesis that tumour CD274 (PD-L1) expression levels might be inversely associated with T-cell densities in colorectal cancer. We found that higher tumour CD274 expression was independently associated with lower density of FOXP3+ cells in human colorectal carcinoma tissue, after controlling for potential confounders, including MSI status, CIMP status and LINE-1 methylation level; these tumour molecular features have been associated with histological lymphocytic reaction and tumour infiltrating T cells in colorectal cancer.4 ,13–15 ,20 Evidence from our current study suggests a possible role of CD274 expression of tumour cells in regulating host immunity in colorectal cancer microenvironment.

    Two previous studies have reported that tumour CD274 expression levels are positively associated with expansion of FOXP3+ regulatory T cells in surgically resected colorectal carcinoma tissue,32 ,33 which is not consistent with our current data. However, these previous studies were severely limited by their small sample sizes (n=33 and n=56), in contrast to our much larger sample size.

    Colorectal cancers are a heterogeneous group of diseases that result from the accumulation of differing sets of genomic and epigenomic alterations and influenced by tumour–host interactions.34–37 The analyses of host immunity against human cancer are increasingly important in cancer research and clinical practice.38 ,39 High densities of CD3+, CD8+ and CD45RO+ cells in colorectal cancer tissue have been associated with better patient survival,14–19 suggesting the densities of these T cells assessed by IHC as a measure of antitumour T-cell-mediated immune response to colorectal tumours.

    High FOXP3+ cell density has been generally associated with favourable outcome of patients with colorectal cancer, although functional roles of FOXP3+ regulatory T cells in various types of cancers remain poorly understood.14 ,17–19 FOXP3+ regulatory T cells, which have been considered as an immunosuppressive subset of T lymphocytes, are functionally and phenotypically diverse, with multiple possible origins and various functional profiles.40 An enhanced infiltration of FOXP3+ cells has been detected in MSI-high colorectal cancer, which is a favourable tumour molecular subtype.15 Accumulating evidence indicates that function of FOXP3+ regulatory T cells can be tailored for differing immune milieu and contexts, and that their roles for cancer progression (tumour-promoting or tumour-suppressive roles) appear to depend on tumour site and progression stage, probably reflecting alterations of the tumour microenvironment.18 ,40 Ladoire et al 41 have proposed that the association of FOXP3+ cells in colorectal cancer tissue with better outcome reflects their capacity of repressing tumour-promoting inflammatory responses to gut microbiota. It is possible that FOXP+ regulatory T cells may have a role in suppressing tumour progression via regulating tumour-promoting inflammation in the colorectal cancer microenvironment.

    We observed CD274 expression in the cytoplasm and membrane of colorectal cancer cells, as many other investigators reported,3 ,32 ,33 ,42–44 with the use of anti-CD274 antibody clone that was referenced and validated in a number of studies for various tumour types.45–48 However, lack of widely accepted immunohistochemical method for FFPE tissue has been a considerable challenge in the assessment of cellular CD274 expression.45 ,49 CD274 expression on the cell surface membrane is likely important for binding to its receptor PDCD1 (PD-1). However, membrane expression was considerably masked in cells with high-level cytoplasmic expression, in contrast to cells with no cytoplasmic expression. Hence, we used both membrane and cytoplasmic expression levels to calculate the tumour CD274 expression score. To assess replicability of our immunohistochemical assessments, we conducted blinded and independent assessment of CD274 expression by the two pathologists, which yielded reasonably good interobserver agreement (a weighted κ of 0.65). Any random misclassification of tumour CD274 expression score would have driven our results towards the null hypothesis. However, despite limitations of immunohistochemical evaluation for CD274 expression, we were able to observe significant inverse association of tumour CD274 expression with FOXP3+ cell density in colorectal cancer tissues.

    Our results suggesting a potential inverse association between tumour MSI-high status and tumour CD274 expression level are intriguing. In contrast, the previous study described that tumour CD274 expression was detected in 5 of 15 MSI-high colorectal cancers and only 1 of 11 MSS cancers3; however, statistical power was severely limited. Another report showed that strong tumour CD274 expression was observed in 433 (36%) of 1197 MSS colorectal cancers, and in 62 (29%) of 223 MSI-high cancers (p=0.040 by the χ2 test),42 consistent with our findings. Another study reported that MSI-high colorectal cancers harboured a larger number of CD274-expressing myeloid cells in tumour tissue than MSS cancers.4 In our current study, we did not observe significant association of the presence of CD274-expressing stromal cells with MSI status in colorectal cancers; however, we found a limited proportion of cases with detectable CD274-expressing stromal cells (5%). Although Herbst et al 50 have underscored the importance of CD274-expressing myeloid cells rather than CD274-expressing tumour cells in the prediction of response to PDCD1 (PD-1) blockade therapy, the relative importance of CD274 expression in stromal cells versus that in tumour cells remains to be clarified.1 Further studies are needed to determine the possible association of tumour MSI status with immune checkpoint molecules in tumour cells and stromal cells.

    We did not observe significant association of tumour CD274 expression levels with colorectal cancer mortality. Previous studies42–45 have reported conflicting results in terms of the association between tumour CD274 expression and clinical outcome in colorectal cancer, possibly due to differences in study populations, designs and methods.

    We observed significant inverse association of tumour CD274 expression with FOXP3+ cell density, but not with CD3+, CD8+ or CD45RO+ cell density, suggesting that influences of tumour CD274 expression on T-cell densities vary by different T-cell subpopulations.

    One limitation of the current study is its cross-sectional design. Hence, we cannot exclude a possibility of reverse causation. It is possible that effects of T cells on tumour cells might change tumour CD274 expression levels. However, our specific hypothesis was based on several lines of experimental evidence indicating that tumour CD274 expression suppresses T-cell-mediated immune response against tumour.1–6 We recognise the limitations in evaluating T cells in human colorectal cancer tissue. We evaluated the well-characterised T-cell markers such as CD3, CD8, CD45RO and FOXP3 with the use of TMA IHC and computer-assisted image analysis to objectively quantify the density of T cells in a large number of cases. In addition, we evaluated up to four tissue cores from each tumour, considering spatial heterogeneity of immune infiltrates. Lack of data on postoperative treatment was another limitation. However, distributions of chemotherapy use and its regimen would unlikely substantially differ according to tumour CD274 expression levels as these data were not available for treatment decision making. Lastly, we admit the limitation of TMA-based assessment, although we did not observe considerable intratumoural heterogeneity in terms of tumour CD274 expression in our validation study using the whole tissue sections. Previous studies suggest that CD274 expression has been prominent (especially in myeloid cells) at the tumour margin of colorectal cancer tissue.3 ,4 Considering spatial tumour heterogeneity, when constructing TMA blocks, we punched tumour area, including tumour centre and tumour margin, to select up to four tumour tissue cores from each tumour. Potential misclassification of tumours in terms of CD274 expression due to intratumoural heterogeneity would be expected to be distributed nearly at random, and hence would have driven our results towards the null hypothesis.

    Strengths of our current study include the use of our molecular pathological epidemiology51 ,52 database of a large number of colorectal cancer cases in the two US nationwide prospective cohort studies. This population-based colorectal cancer database enabled us to rigorously examine the association of tumour CD274 expression with histological lymphocytic reaction or with the density of T cells, controlling for potential confounders. In addition, our colorectal cancer specimens were derived from a large number of hospitals in diverse settings across the USA (but not based on a limited number of hospitals), which increase the generalisability of our findings.

    In conclusion, our current study has shown that tumour CD274 expression level is inversely associated with the density of FOXP3+ lymphocytes in colorectal carcinoma tissue. Upon validation, our human population data suggest an influence of tumour CD274 expression on regulatory T cells, and can inform further mechanistic studies to elucidate potential interactive roles of the immune checkpoint pathway and host immunity in colorectal carcinogenesis.

    Acknowledgments

    The authors would like to thank the participants and staff of the Nurses' Health Study and the Health Professionals Follow-up Study for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data.

    References

    1. ↵
      2. Topalian SL ,
      3. Drake CG ,
      4. Pardoll DM
      . Immune checkpoint blockade: a common denominator approach to cancer therapy. Cancer Cell 2015;27:450–61. doi:10.1016/j.ccell.2015.03.001
      OpenUrlCrossRefPubMed
    2. ↵
      2. Tumeh PC ,
      3. Harview CL ,
      4. Yearley JH , et al
      . PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 2014;515:568–71. doi:10.1038/nature13954
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      2. Le DT ,
      3. Uram JN ,
      4. Wang H , et al
      . PD-1 Blockade in tumors with mismatch-repair deficiency. N Engl J Med 2015;372:2509–20. doi:10.1056/NEJMoa1500596
      OpenUrlCrossRefPubMed
    4. ↵
      2. Llosa NJ ,
      3. Cruise M ,
      4. Tam A , et al
      . The vigorous immune microenvironment of microsatellite instable colon cancer is balanced by multiple counter-inhibitory checkpoints. Cancer Discov 2015;5:43–51. doi:10.1158/2159-8290.CD-14-0863
      OpenUrlAbstract/FREE Full Text
    5. ↵
      2. Gentles AJ ,
      3. Newman AM ,
      4. Liu CL , et al
      . The prognostic landscape of genes and infiltrating immune cells across human cancers. Nat Med 2015;21:938–45. doi:10.1038/nm.3909
      OpenUrlCrossRefPubMed
    6. ↵
      2. Lal N ,
      3. Beggs AD ,
      4. Willcox BE , et al
      . An immunogenomic stratification of colorectal cancer: implications for development of targeted immunotherapy. Oncoimmunology 2015;4:e976052. doi:10.4161/2162402X.2014.976052
      OpenUrlCrossRefPubMed
    7. ↵
      2. Okugawa Y ,
      3. Grady WM ,
      4. Goel A
      . Epigenetic alterations in colorectal cancer: emerging biomarkers. Gastroenterology 2015;149:1204–25.e12. doi:10.1053/j.gastro.2015.07.011
      OpenUrlCrossRefPubMed
    8. ↵
      2. Colussi D ,
      3. Brandi G ,
      4. Bazzoli F , et al
      . Molecular pathways involved in colorectal cancer: implications for disease behavior and prevention. Int J Mol Sci 2013;14:16365–85. doi:10.3390/ijms140816365
      OpenUrlCrossRefPubMed
    9. ↵
      2. Di Caro G ,
      3. Marchesi F ,
      4. Laghi L , et al
      . Immune cells: plastic players along colorectal cancer progression. J Cell Mol Med 2013;17:1088–95. doi:10.1111/jcmm.12117
      OpenUrlCrossRefPubMed
    10. ↵
      2. Dienstmann R ,
      3. Salazar R ,
      4. Tabernero J
      . Personalizing colon cancer adjuvant therapy: selecting optimal treatments for individual patients. J Clin Oncol 2015;33:1787–96. doi:10.1200/JCO.2014.60.0213
      OpenUrlAbstract/FREE Full Text
    11. ↵
      2. Ijspeert JE ,
      3. Vermeulen L ,
      4. Meijer GA , et al
      . Serrated neoplasia-role in colorectal carcinogenesis and clinical implications. Nat Rev Gastroenterol Hepatol 2015;12:401–9. doi:10.1038/nrgastro.2015.73
      OpenUrlCrossRefPubMed
    12. ↵
      2. Kuipers EJ ,
      3. Grady WM ,
      4. Lieberman D , et al
      . Colorectal cancer. Nat Rev Dis Primers 2015. doi:10.1038/nrdp.2015.65
    13. ↵
      2. Ogino S ,
      3. Galon J ,
      4. Fuchs CS , et al
      . Cancer immunology—analysis of host and tumor factors for personalized medicine. Nat Rev Clin Oncol 2011;8:711–19. doi:10.1038/nrclinonc.2011.122
      OpenUrlCrossRefPubMed
    14. ↵
      2. Palmqvist R ,
      3. Wikberg M ,
      4. Ling A , et al
      . The association of immune cell infiltration and prognosis in colorectal cancer. Curr Colorectal Cancer Rep 2013;9:372–9. doi:10.1007/s11888-013-0192-3
      OpenUrlCrossRef
    15. ↵
      2. Nosho K ,
      3. Baba Y ,
      4. Tanaka N , et al
      . Tumour-infiltrating T-cell subsets, molecular changes in colorectal cancer, and prognosis: cohort study and literature review. J Pathol 2010;222:350–66. doi:10.1002/path.2774
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      2. Galon J ,
      3. Costes A ,
      4. Sanchez-Cabo F , et al
      . Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 2006;313:1960–4. doi:10.1126/science.1129139
      OpenUrlAbstract/FREE Full Text
    17. ↵
      2. Mlecnik B ,
      3. Bindea G ,
      4. Angell HK , et al
      . Functional network pipeline reveals genetic determinants associated with in situ lymphocyte proliferation and survival of cancer patients. Sci Transl Med 2014;6:228ra37. doi:10.1126/scitranslmed.3007240
      OpenUrlAbstract/FREE Full Text
    18. ↵
      2. Shang B ,
      3. Liu Y ,
      4. Jiang SJ , et al
      . Prognostic value of tumor-infiltrating FoxP3+ regulatory T cells in cancers: a systematic review and meta-analysis. Sci Rep 2015;5:15179. doi:10.1038/srep15179
      OpenUrlCrossRefPubMed
    19. ↵
      2. Mei Z ,
      3. Liu Y ,
      4. Liu C , et al
      . Tumour-infiltrating inflammation and prognosis in colorectal cancer: systematic review and meta-analysis. Br J Cancer 2014;110:1595–605. doi:10.1038/bjc.2014.46
      OpenUrlCrossRefPubMed
    20. ↵
      2. Ogino S ,
      3. Nosho K ,
      4. Irahara N , et al
      . Lymphocytic reaction to colorectal cancer is associated with longer survival, independent of lymph node count, microsatellite instability, and CpG island methylator phenotype. Clin Cancer Res 2009;15:6412–20. doi:10.1158/1078-0432.CCR-09-1438
      OpenUrlAbstract/FREE Full Text
    21. ↵
      2. Nishihara R ,
      3. Lochhead P ,
      4. Kuchiba A , et al
      . Aspirin use and risk of colorectal cancer according to BRAF mutation status. JAMA 2013;309:2563–71. doi:10.1001/jama.2013.6599
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Liao X ,
      3. Lochhead P ,
      4. Nishihara R , et al
      . Aspirin use, tumor PIK3CA mutation, and colorectal-cancer survival. N Engl J Med 2012;367:1596–606. doi:10.1056/NEJMoa1207756
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      2. Nishihara R ,
      3. Wu K ,
      4. Lochhead P , et al
      . Long-term colorectal-cancer incidence and mortality after lower endoscopy. N Engl J Med 2013;369:1095–105. doi:10.1056/NEJMoa1301969
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      2. Yamauchi M ,
      3. Lochhead P ,
      4. Morikawa T , et al
      . Colorectal cancer: a tale of two sides or a continuum? Gut 2012;61:794–7. doi:10.1136/gutjnl-2012-302014
      OpenUrlFREE Full Text
    25. ↵
      2. Chan AT ,
      3. Ogino S ,
      4. Fuchs CS
      . Aspirin and the risk of colorectal cancer in relation to the expression of COX-2. N Engl J Med 2007;356:2131–42. doi:10.1056/NEJMoa067208
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      2. Ogino S ,
      3. Nosho K ,
      4. Kirkner GJ , et al
      . CpG island methylator phenotype, microsatellite instability, BRAF mutation and clinical outcome in colon cancer. Gut 2009;58:90–6. doi:10.1136/gut.2008.155473
      OpenUrlAbstract/FREE Full Text
    27. ↵
      2. Nosho K ,
      3. Irahara N ,
      4. Shima K , et al
      . Comprehensive biostatistical analysis of CpG island methylator phenotype in colorectal cancer using a large population-based sample. PLoS ONE 2008;3:e3698. doi:10.1371/journal.pone.0003698
      OpenUrlCrossRefPubMed
    28. ↵
      2. Imamura Y ,
      3. Lochhead P ,
      4. Yamauchi M , et al
      . Analyses of clinicopathological, molecular, and prognostic associations of KRAS codon 61 and codon 146 mutations in colorectal cancer: cohort study and literature review. Mol Cancer 2014;13:135. doi:10.1186/1476-4598-13-135
      OpenUrlCrossRefPubMed
    29. ↵
      2. Liao X ,
      3. Morikawa T ,
      4. Lochhead P , et al
      . Prognostic role of PIK3CA mutation in colorectal cancer: cohort study and literature review. Clin Cancer Res 2012;18:2257–68. doi:10.1158/1078-0432.CCR-11-2410
      OpenUrlAbstract/FREE Full Text
    30. ↵
      2. Mima K ,
      3. Sukawa Y ,
      4. Nishihara R , et al
      . Fusobacterium nucleatum and T cells in colorectal carcinoma. JAMA Oncol 2015;1:653–61. doi:10.1001/jamaoncol.2015.1377
      OpenUrl
    31. ↵
      2. Mima K ,
      3. Nishihara R ,
      4. Qian ZR , et al
      . Fusobacterium nucleatum in colorectal carcinoma tissue and patient prognosis. Gut Published Online First: 26 Aug 2015. doi:10.1136/gutjnl-2015-310101 doi:10.1136/gutjnl-2015-310101
    32. ↵
      2. Zhao LW ,
      3. Li C ,
      4. Zhang RL , et al
      . B7-H1 and B7-H4 expression in colorectal carcinoma: correlation with tumor FOXP3(+) regulatory T-cell infiltration. Acta Histochem 2014;116:1163–8. doi:10.1016/j.acthis.2014.06.003
      OpenUrlCrossRefPubMed
    33. ↵
      2. Hua D ,
      3. Sun J ,
      4. Mao Y , et al
      . B7-H1 expression is associated with expansion of regulatory T cells in colorectal carcinoma. World J Gastroenterol 2012;18:971–8. doi:10.3748/wjg.v18.i9.971
      OpenUrlCrossRefPubMed
    34. ↵
      2. Kocarnik JM ,
      3. Shiovitz S ,
      4. Phipps AI
      . Molecular phenotypes of colorectal cancer and potential clinical applications. Gastroenterol Rep (Oxf) 2015;3:269–76. doi:10.1093/gastro/gov046
      OpenUrlAbstract/FREE Full Text
    35. ↵
      2. Ogino S ,
      3. Fuchs CS ,
      4. Giovannucci E
      . How many molecular subtypes? Implications of the unique tumor principle in personalized medicine. Expert Rev Mol Diagn 2012;12:621–8. doi:10.1586/erm.12.46
      OpenUrlCrossRefPubMed
    36. ↵
      2. Martinez-Useros J ,
      3. Garcia-Foncillas J
      . Obesity and colorectal cancer: molecular features of adipose tissue. J Transl Med 2016;14:21. doi:10.1186/s12967-016-0772-5
      OpenUrlCrossRefPubMed
    37. ↵
      2. Roy HK ,
      3. Turzhitsky V ,
      4. Wali R , et al
      . Spectral biomarkers for chemoprevention of colonic neoplasia: a placebo-controlled double-blinded trial with aspirin. Gut Published Online First: 26 Oct 2015. doi:10.1136/gutjnl-2015-309996
    38. ↵
      2. Galon J ,
      3. Mlecnik B ,
      4. Bindea G , et al
      . Towards the introduction of the ‘Immunoscore’ in the classification of malignant tumours. J Pathol 2014;232:199–209. doi:10.1002/path.4287
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      2. Mlecnik B ,
      3. Bindea G ,
      4. Angell HK , et al
      . Integrative analyses of colorectal cancer show immunoscore is a stronger predictor of patient survival than microsatellite instability. Immunity 2016;44:698–711. doi:10.1016/j.immuni.2016.02.025
      OpenUrlCrossRefPubMed
    40. ↵
      2. Liston A ,
      3. Gray DH
      . Homeostatic control of regulatory T cell diversity. Nat Rev Immunol 2014;14:154–65. doi:10.1038/nri3605
      OpenUrlCrossRefPubMed
    41. ↵
      2. Ladoire S ,
      3. Martin F ,
      4. Ghiringhelli F
      . Prognostic role of FOXP3+ regulatory T cells infiltrating human carcinomas: the paradox of colorectal cancer. Cancer Immunol Immunother 2011;60:909–18. doi:10.1007/s00262-011-1046-y
      OpenUrlCrossRefPubMedWeb of Science
    42. ↵
      2. Droeser RA ,
      3. Hirt C ,
      4. Viehl CT , et al
      . Clinical impact of programmed cell death ligand 1 expression in colorectal cancer. Eur J Cancer 2013;49:2233–42. doi:10.1016/j.ejca.2013.02.015
      OpenUrlCrossRefPubMed
    43. ↵
      2. Liang M ,
      3. Li J ,
      4. Wang D , et al
      . T-cell infiltration and expressions of T lymphocyte co-inhibitory B7-H1 and B7-H4 molecules among colorectal cancer patients in northeast China's Heilongjiang province. Tumour Biol 2014;35:55–60. doi:10.1007/s13277-013-1006-6
      OpenUrlCrossRefPubMed
    44. ↵
      2. Song M ,
      3. Chen D ,
      4. Lu B , et al
      . PTEN loss increases PD-L1 protein expression and affects the correlation between PD-L1 expression and clinical parameters in colorectal cancer. PLoS ONE 2013;8:e65821. doi:10.1371/journal.pone.0065821
      OpenUrlCrossRefPubMed
    45. ↵
      2. Wu P ,
      3. Wu D ,
      4. Li L , et al
      . PD-L1 and survival in solid tumors: a meta-analysis. PLoS ONE 2015;10:e0131403. doi:10.1371/journal.pone.0131403
      OpenUrlCrossRefPubMed
    46. ↵
      2. Ohigashi Y ,
      3. Sho M ,
      4. Yamada Y , et al
      . Clinical significance of programmed death-1 ligand-1 and programmed death-1 ligand-2 expression in human esophageal cancer. Clin Cancer Res 2005;11:2947–53. doi:10.1158/1078-0432.CCR-04-1469
      OpenUrlAbstract/FREE Full Text
    47. ↵
      2. Nomi T ,
      3. Sho M ,
      4. Akahori T , et al
      . Clinical significance and therapeutic potential of the programmed death-1 ligand/programmed death-1 pathway in human pancreatic cancer. Clin Cancer Res 2007;13:2151–7. doi:10.1158/1078-0432.CCR-06-2746
      OpenUrlAbstract/FREE Full Text
    48. ↵
      2. Youngnak P ,
      3. Kozono Y ,
      4. Kozono H , et al
      . Differential binding properties of B7-H1 and B7-DC to programmed death-1. Biochem Biophys Res Commun 2003;307:672–7. doi:10.1016/S0006-291X(03)01257-9
      OpenUrlCrossRefPubMedWeb of Science
    49. ↵
      2. Taube JM ,
      3. Klein A ,
      4. Brahmer JR , et al
      . Association of PD-1, PD-1 ligands, and other features of the tumor immune microenvironment with response to anti-PD-1 therapy. Clin Cancer Res 2014;20:5064–74. doi:10.1158/1078-0432.CCR-13-3271
      OpenUrlAbstract/FREE Full Text
    50. ↵
      2. Herbst RS ,
      3. Soria JC ,
      4. Kowanetz M , et al
      . Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients. Nature 2014;515:563–7. doi:10.1038/nature14011
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      2. Ogino S ,
      3. Chan AT ,
      4. Fuchs CS , et al
      . Molecular pathological epidemiology of colorectal neoplasia: an emerging transdisciplinary and interdisciplinary field. Gut 2011;60:397–411. doi:10.1136/gut.2010.217182
      OpenUrlAbstract/FREE Full Text
    52. ↵
      2. Ogino S ,
      3. Nishihara R ,
      4. VanderWeele TJ , et al
      . The role of molecular pathological epidemiology in the study of neoplastic and non-neoplastic diseases in the era of precision medicine. Epidemiology 2016 (in press). doi:10.1097/EDE.0000000000000471
    View Abstract

    Footnotes

    • YM, RN, JY, KM, SR, CSF, ZRQ and SO contributed equally.

    • Contributors All authors contributed to review and revision. CSF, ZRQ and SO: developed the main concept and designed the study. ATC, CSF and SO: wrote grant applications. YM, RN, JY, KM, AdS, YS, KI, YC, MS, JAN, XL, KN, ATC, CSF, ZRQ and SO: were responsible for collection of tumour tissue, and acquisition of epidemiologic, clinical and tumour tissue data, including histopathological and immunohistochemical characteristics. YM, RN, KM, CSF, ZRQ and SO: performed data analysis and interpretation. YM, KM and SO: drafted the manuscript. RN, JY, KM, YC, MG, AJB, FSH, GJF, SR, CSF, ZRQ and SO: contributed to editing and critical revision for important intellectual contents.

    • Funding This work was supported by US National Institutes of Health (NIH) grants (P01 CA87969 to MJ Stampfer; UM1 CA186107 to MJ Stampfer; P01 CA55075 to WC Willett; UM1 CA167552 to WC Willett; P50 CA127003 to CSF; R01 CA137178 to ATC; R01 CA151993 to SO; R35 CA197735 to SO and K07 CA190673 to RN); and by grants from the Project P Fund, the Friends of the Dana-Farber Cancer Institute, Bennett Family Fund and the Entertainment Industry Foundation through National Colorectal Cancer Research Alliance. KM is supported by a grant from Program for Advancing Strategic International Networks to Accelerate the Circulation of Talented Researchers from Japanese Society for the Promotion of Science. YM is supported by a fellowship grant of the Keio Gijuku Fukuzawa Memorial Fund for the Advancement of Education and Research. JY is supported by National Natural Science Foundation of China (81200612).

    • Competing interests ATC previously served as a consultant for Bayer Healthcare, Millennium Pharmaceuticals, Pozen and Pfizer. FSH has served as a consultant to Merck, Novartis and Genentech and has received grant support to institution from Bristol-Myers Squibb.

    • Ethics approval The Institutional Review Boards at the Harvard T.H. Chan School of Public Health and the Brigham and Women’s Hospital (Boston, Massachusetts, USA).

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

    • Data sharing statement With input from and approval of the External Advisory Committees for the two cohorts (the Nurses’ Health Study and the Health Professionals Follow-up Study), we have adopted a data enclave approach to data sharing. Along with a general description of the Nurses’ Health Study and Health Professionals Follow-up Study cohorts as well as all questionnaires that have been used since 1976 for both the blood repository and the questionnaire data, guidelines are available on our website for outside users to access the resources of our study (http://www.hsph.harvard.edu/hpfs/). Typically, an outside user prepares a brief proposal that is reviewed by the Nurses Health Study and Health Professionals Follow-up Study investigator groups to identify a local investigator to assist the outside investigator. If uncertainty arises as to the merit of a request, the External Advisory Committee makes a final decision.