Immune Profiling of Deficient Mismatch Repair Colorectal Cancer Tumor Microenvironment Reveals Different Levels of Immune System Activation
2020; Elsevier BV; Volume: 22; Issue: 5 Linguagem: Inglês
10.1016/j.jmoldx.2020.02.008
ISSN1943-7811
AutoresRiccardo Giannini, Gemma Zucchelli, Mirella Giordano, Clara Ugolini, Roberto Moretto, Katarzyna E. Ambryszewska, Michele Leonardi, Elisa Sensi, Federica Morano, Filippo Pietrantonio, Chiara Cremolini, Alfredo Falcone, Gabriella Fontanini,
Tópico(s)Ferroptosis and cancer prognosis
ResumoTo understand the immune landscape of deficient mismatch repair colorectal cancer (dMMR CRC) tumor microenvironment, gene expression profiling was performed by the nCounter PanCancer Immune Profiling Panel. This study was conducted retrospectively on 89 dMMR-CRC samples. The expression of CD3, CD8, programmed death-1, and programmed death ligand-1 protein was evaluated on a subset of samples by immunohistochemistry, and lymphocyte density was calculated. A subset of deregulated genes was identified. Functional clustering analysis performed on these genes generated four main factors: antigen processing and presentation, with its major histocompatibility complex-II–related genes; genes correlated with the cytotoxic activity of immune system; T-cell chemotaxis/cell adhesion genes; and T-CD4+ regulator cell–related genes. A deregulation score (DS) was calculated for each sample. On the basis of their DS, tumors were then classified as COLD (DS ≤ −3) to select the samples with a strong down-regulation of the immune system and NOT COLD (DS ≥ −2). The COLD group of patients showed a worse prognosis in terms of survival considering all patients (P = 0.0172) and patients with metastatic disease (P = 0.0031). These results confirm that dMMR-CRCs do not constitute a homogeneous group as concerns the immune system activity of tumor microenvironment. In particular, the distinction between COLD and NOT COLD tumors may improve the management of these two subsets of patients. To understand the immune landscape of deficient mismatch repair colorectal cancer (dMMR CRC) tumor microenvironment, gene expression profiling was performed by the nCounter PanCancer Immune Profiling Panel. This study was conducted retrospectively on 89 dMMR-CRC samples. The expression of CD3, CD8, programmed death-1, and programmed death ligand-1 protein was evaluated on a subset of samples by immunohistochemistry, and lymphocyte density was calculated. A subset of deregulated genes was identified. Functional clustering analysis performed on these genes generated four main factors: antigen processing and presentation, with its major histocompatibility complex-II–related genes; genes correlated with the cytotoxic activity of immune system; T-cell chemotaxis/cell adhesion genes; and T-CD4+ regulator cell–related genes. A deregulation score (DS) was calculated for each sample. On the basis of their DS, tumors were then classified as COLD (DS ≤ −3) to select the samples with a strong down-regulation of the immune system and NOT COLD (DS ≥ −2). The COLD group of patients showed a worse prognosis in terms of survival considering all patients (P = 0.0172) and patients with metastatic disease (P = 0.0031). These results confirm that dMMR-CRCs do not constitute a homogeneous group as concerns the immune system activity of tumor microenvironment. In particular, the distinction between COLD and NOT COLD tumors may improve the management of these two subsets of patients. The defective function of the mismatch repair (MMR) system causes the accumulation of gene mutations, such as insertions or deletions that induce an increase of neoantigens and elicit a remarkable endogenous antitumor immune response. Microsatellite instability (MSI) is a phenotypic marker of the deficient DNA MMR system and drives the pathogenesis of approximately 15% of colorectal cancers (CRCs). The system is composed by four proteins (DNA mismatch repair protein Mlh1, DNA mismatch repair protein Msh2, DNA mismatch repair protein Msh6, and mismatch repair endonuclease PMS2) and is involved in the repair of DNA sequence mismatches that occur during DNA replication.1Gelsomino F. Barbolini M. Spallanzani A. Pugliese G. Cascinu S. The evolving role of microsatellite instability in colorectal cancer: a review.Cancer Treat Rev. 2016; 51: 19-26Abstract Full Text Full Text PDF PubMed Scopus (156) Google Scholar The active immune microenvironment is counterbalanced by the strong expression of immunosuppressive ligands and signals, including programmed death-1 (PD-1) and programmed death ligand-1 (PD-L1). For these reasons, MSI tumors represent an ideal model to assess the activity and efficacy of checkpoint inhibitors. Indeed, pembrolizumab and nivolumab have shown impressive response rates and long-lasting survival in chemorefractory MSI metastatic CRC patients.2Le D.T. Durham J.N. Smith K.N. Wang H. Bartlett B.R. Aulakh L.K. et al.Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade.Science. 2017; 357: 409-413Crossref PubMed Scopus (3768) Google Scholar, 3Le D.T. Uram J.N. Wang H. Bartlett B.R. Kemberling H. Eyring A.D. Skora A.D. Luber B.S. Azad N.S. Laheru D. Biedrzycki B. Donehower R.C. Zaheer A. Fisher G.A. Crocenzi T.S. Lee J.J. Duffy S.M. Goldberg R.M. de la Chapelle A. Koshiji M. Bhaijee F. Huebner T. Hruban R.H. Wood L.D. Cuka N. Pardoll D.M. Papadopoulos N. Kinzler K.W. Zhou S. Cornish T.C. Taube J.M. Anders R.A. Eshleman J.R. Vogelstein B. Diaz L.A. PD-1 blockade in tumors with mismatch-repair deficiency.N Engl J Med. 2015; 372: 2509-2520Crossref PubMed Scopus (6107) Google Scholar, 4Overman M.J. Lonardi S. Wong K.Y.M. Lenz H.-J. Gelsomino F. Aglietta M. Morse M.A. Van Cutsem E. McDermott R. Hill A. Sawyer M.B. Hendlisz A. Neyns B. Svrcek M. Moss R.A. Ledeine J.-M. Cao Z.A. Kamble S. Kopetz S. André T. Durable clinical benefit with nivolumab plus ipilimumab in DNA mismatch repair-deficient/microsatellite instability-high metastatic colorectal cancer.J Clin Oncol. 2018; 36: 773-779Crossref PubMed Scopus (1096) Google Scholar, 5Overman M.J. McDermott R. Leach J.L. Lonardi S. Lenz H.-J. Morse M.A. Desai J. Hill A. Axelson M. Moss R.A. Goldberg M.V. Cao Z.A. Ledeine J.-M. Maglinte G.A. Kopetz S. André T. Nivolumab in patients with metastatic DNA mismatch repair-deficient or microsatellite instability-high colorectal cancer (CheckMate 142): an open-label, multicentre, phase 2 study.Lancet Oncol. 2017; 18: 1182-1191Abstract Full Text Full Text PDF PubMed Scopus (1536) Google Scholar However, the studies assessing checkpoint blockade immunotherapies in MSI metastatic CRC have demonstrated that only half of these patients benefited from anti–PD-1 inhibitors, despite the strong biological rationale.3Le D.T. Uram J.N. Wang H. Bartlett B.R. Kemberling H. Eyring A.D. Skora A.D. Luber B.S. Azad N.S. Laheru D. Biedrzycki B. Donehower R.C. Zaheer A. Fisher G.A. Crocenzi T.S. Lee J.J. Duffy S.M. Goldberg R.M. de la Chapelle A. Koshiji M. Bhaijee F. Huebner T. Hruban R.H. Wood L.D. Cuka N. Pardoll D.M. Papadopoulos N. Kinzler K.W. Zhou S. Cornish T.C. Taube J.M. Anders R.A. Eshleman J.R. Vogelstein B. Diaz L.A. PD-1 blockade in tumors with mismatch-repair deficiency.N Engl J Med. 2015; 372: 2509-2520Crossref PubMed Scopus (6107) Google Scholar,5Overman M.J. McDermott R. Leach J.L. Lonardi S. Lenz H.-J. Morse M.A. Desai J. Hill A. Axelson M. Moss R.A. Goldberg M.V. Cao Z.A. Ledeine J.-M. Maglinte G.A. Kopetz S. André T. Nivolumab in patients with metastatic DNA mismatch repair-deficient or microsatellite instability-high colorectal cancer (CheckMate 142): an open-label, multicentre, phase 2 study.Lancet Oncol. 2017; 18: 1182-1191Abstract Full Text Full Text PDF PubMed Scopus (1536) Google Scholar One possible explanation for this different efficacy of checkpoint inhibitors could be because of the immune-microenvironment heterogeneity of MSI CRCs. In other cancer types, such as melanoma, which present high mutational burden, gene expression profiling has allowed to detect two major phenotypes of tumor microenvironment: a T-cell–inflamed phenotype, characterized by the expression of T-lymphocyte markers and chemokines correlated with the recruitment of T lymphocytes; and a non–T-cell–inflamed phenotype missing the expression of immune-related genes. Typically, the T-cell–inflamed phenotype is also characterized by high representation of immune-inhibitory factors, including expression of the membrane protein PD-L1, expression of the tryptophan-catabolizing enzyme indoleamine 2,3-dioxygenase 1, and infiltration of forkhead box protein P3–positive T-regulatory (Treg) lymphocytes, which indicate the occurrence of immune escape in the context of an antitumor immune response.6Schreiber R.D. Old L.J. Smyth M.J. Cancer immunoediting: integrating immunity's roles in cancer suppression and promotion.Science. 2011; 331: 1565-1570Crossref PubMed Scopus (4070) Google Scholar, 7Dunn G.P. Old L.J. Schreiber R.D. The immunobiology of cancer immunosurveillance and immunoediting.Immunity. 2004; 21: 137-148Abstract Full Text Full Text PDF PubMed Scopus (2114) Google Scholar, 8Vesely M.D. Kershaw M.H. Schreiber R.D. Smyth M.J. Natural innate and adaptive immunity to cancer.Annu Rev Immunol. 2011; 29: 235-271Crossref PubMed Scopus (1431) Google Scholar, 9Trujillo J.A. Sweis R.F. Bao R. Luke J.J. T cell-inflamed versus non-T cell-inflamed tumors: a conceptual framework for cancer immunotherapy drug development and combination therapy selection.Cancer Immunol Res. 2018; 6: 990-1000Crossref PubMed Scopus (221) Google Scholar, 10Boland P.M. Ma W.W. Immunotherapy for colorectal cancer.Cancers. 2017; 9: E50Crossref PubMed Scopus (102) Google Scholar, 11Spranger S. Luke J.J. Bao R. Zha Y. Hernandez K.M. Li Y. Gajewski A.P. Andrade J. Gajewski T.F. Density of immunogenic antigens does not explain the presence or absence of the T-cell-inflamed tumor microenvironment in melanoma.Proc Natl Acad Sci U S A. 2016; 113: E7759-E7768Crossref PubMed Scopus (254) Google Scholar, 12Gajewski T.F. Corrales L. Williams J. Horton B. Sivan A. Spranger S. Cancer immunotherapy targets based on understanding the T cell-inflamed versus non-T cell-inflamed tumor microenvironment.Adv Exp Med Biol. 2017; 1036: 19-31Crossref PubMed Scopus (156) Google Scholar Patients presenting a T-cell–inflamed immune phenotype respond better to different immunotherapeutic approaches, such as anticancer vaccines, high-dose IL-2, and inhibitory antibodies directed against cytotoxic T-lymphocyte associated-4, PD-1, and PD-L1.13Fridman W.H. Pagès F. Sautès-Fridman C. Galon J. The immune contexture in human tumours: impact on clinical outcome.Nat Rev Cancer. 2012; 12: 298-306Crossref PubMed Scopus (3105) Google Scholar Preclinical studies and in vivo analysis of specific biomarkers have suggested that the therapeutic activity of these immunotherapies is associated with the reactivation in the tumor microenvironment of T lymphocytes capable of recognizing tumor antigens.14Melero I. Rouzaut A. Motz G.T. Coukos G. T-cell and NK-cell infiltration into solid tumors: a key limiting factor for efficacious cancer immunotherapy.Cancer Discov. 2014; 4: 522-526Crossref PubMed Scopus (262) Google Scholar On the basis of these considerations, an mRNA expression analysis was performed to describe the immune profile of deficient MMR (dMMR) CRCs. Primary tumor samples [formalin fixed, paraffin embedded (FFPE)] of 89 dMMR CRC patients referred to two Italian Oncology Units (Azienda Ospedaliero–Universitaria Pisana, Pisa, Italy; and National Cancer Institute, Milan, Italy) were collected from 2012 to 2017. CRCs were selected according to the absence of the protein expression encoded by the corresponding MMR genes (hMLH1, hMSH2, hMSH6, or PMS2), which reflects the MSI phenotype. The most representative paraffin block of each sample was selected for analysis with necrosis areas and regression zones excluded. The study was conducted anonymously and in compliance with the principles of the Declaration of Helsinki of 1975. For each sample, four unstained sections (5 μm thick) were used for RNA extraction. The unstained sections were deparaffinized with xylene and rehydrated in decreasing-grade ethanol solution. Manual microdissection was performed to maximize the amount of tumor cells. RNA was isolated by using the RNeasy FFPE Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. RNA was eluted in 20 μL of RNase-free water. RNA quantity and quality were assessed by means of a spectrophotometer (Xpose Trinean, Gentbrugge, Belgium). Analysis of the expression profiles of >700 immune-related genes was performed by the nanoString nCounter PanCancer Immune Profiling Panel (NanoString Technologies, Seattle, WA). In detail, 150 ng of RNA from each sample was hybridized with the nCounter PanCancer Immune Profiling Panel (GX Assay) CodeSet. All the procedures related to mRNA quantification, including sample preparation, hybridization, detection, and scanning, were performed following the manufacturer's instructions. The counts were normalized according to the standard protocol. Raw nanoString counts for each mRNA within each experiment were subjected to technical normalization with the counts obtained for positive-control probe sets before biological normalization using the 40 reference genes included in the CodeSet. The normalized data were log2 transformed and then used as input for differential expression analysis. The data were filtered to exclude relatively invariant features and features below the detection threshold (defined for each sample by a cutoff value corresponding to twice the SD of the negative control probes plus the means). The PanCancer Immune Profiling Advanced Analysis Module (NanoString Technologies) was used to conduct the statistical analyses of data obtained by the nCounter panel analysis. The analysis module grouped the genes into functional immune-related categories (namely, transporter functions, tumor necrosis factor superfamily, macrophage functions, antigen processing, adhesion, regulation, T-cell functions, cytokines, B-cell functions, ILs, toll-like receptor, cytotoxicity, pathogen defense, cancer/testis antigens, complement, natural killer (NK) cell functions, cell functions, chemokines, leukocyte functions, cell cycle, senescence, and microglial functions). To understand what the immune cell profiling results represented, a set of genes for each cell population was assumed to be specific (reference genes) to that cell type. This assumption allowed measuring the abundance of a cell type by simply taking the average log2 expression of its characteristic genes. This approach was used to test the relative abundance of B cells, T cells (helper T cells, Tregs, cytotoxic cells, CD8+ T cells, and CD45), natural killer cells, dendritic cells, macrophages, mast cells, and neutrophils.15Bindea G. Mlecnik B. Tosolini M. Kirilovsky A. Waldner M. Obenauf A.C. Angell H. Fredriksen T. Lafontaine L. Berger A. Bruneval P. Fridman W.H. Becker C. Pagès F. Speicher M.R. Trajanoski Z. Galon J. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer.Immunity. 2013; 39: 782-795Abstract Full Text Full Text PDF PubMed Scopus (1949) Google Scholar,16Newman A.M. Liu C.L. Green M.R. Gentles A.J. Feng W. Xu Y. Hoang C.D. Diehn M. Alizadeh A.A. Robust enumeration of cell subsets from tissue expression profiles.Nat Methods. 2015; 12: 453-457Crossref PubMed Scopus (5000) Google Scholar To investigate the differential expression between the samples in this study, the main covariates considered were tumor size, tumor stage, and presence of metastases and molecular profile of the samples. The large number of genes in the CodeSet made the use of raw P values problematic. Therefore, the Benjamini-Yekutieli false discovery rate correction was used for adjusting P values; genes with a fold change ≥ 1 and a Benjamini-Yekutieli false discovery rate < 0.05 were considered differentially expressed. Functional clustering analysis was performed starting from the immune-related categories of genes revealed by PanCancer Immune Profiling Advanced Analysis to confirm, and relates to the immune cell population in the observed clusters. Analysis was performed by DAVID Bioinformatics Resources 6.8 and STRING version 10.5.17Szklarczyk D. Morris J.H. Cook H. Kuhn M. Wyder S. Simonovic M. Santos A. Doncheva N.T. Roth A. Bork P. Jensen L.J. von Mering C. The STRING database in 2017: quality-controlled protein–protein association networks, made broadly accessible.Nucleic Acids Res. 2017; 45: D362-D368Crossref PubMed Scopus (4325) Google Scholar to evaluate the gene expression and the corresponding potential protein networks. The Markov Cluster algorithm with inflation = 3 was used to group the genes into annotation clusters on the basis of precomputed similarity information.18Enright A.J. Van Dongen S. Ouzounis C.A. An efficient algorithm for large-scale detection of protein families.Nucleic Acids Res. 2002; 30: 1575-1584Crossref PubMed Scopus (2522) Google Scholar In addition, vector quantization (k-means clustering) with applied number of clusters was revealed. The resulting clusters were annotated and ranked by linear P value (Pearson) with Bonferroni's correction using a threshold of counted genes. Preliminary basic clusters (P < 0.05) were selected for further analysis to determine a strong gene signature correlated with the different immune cell population and their mechanism of action. R-linear correlation analysis of the gene expression levels of the singular clusters with Bonferroni's correction was performed in Past software version 3.21 (https://folk.uio.no/ohammer/past, last accessed October 1, 2018), generating four subsets with strongly correlated genes belonging to the same functional clusters. A refinement and reclassification of the new functional cluster's groups generated was performed by the association of the immune cell types and the corresponding functions generating the final determinant factors. A sample classification was performed using the quartile clustering technique on the basis of the factors determined by functional clustering analysis and gene expression values (from nCounter).19Goswami S. Chakrabarti A. Quartile clustering: a quartile based technique for generating meaningful clusters.J Comput. 2012; 4: 48-55Google Scholar,20Janowitz M.F. Schweizer B. Ordinal and percentile clustering.Math Soc Sci. 1989; 18: 135-186Crossref Scopus (15) Google Scholar Quartiles are a major tool in descriptive analysis, which divides the range of data into three parts. Once the values of upper quartile, median quartile, and lower quartile had been determined for a singular clustered gene, the discrete value was assigned according to the affiliation of the experimental value of gene expression to the specific interquartile range. Calculation of the total score for each of the functional clusters was based on the algebraic sum of the partial grades (∑g) and was assigned in a discrete way using the following condition: −1 if ∑g < 0; 0 if ∑g = 0; and +1 if ∑g > 0. Data were analyzed using both parametric and nonparametric tests using Statistica software version 12 (StatSoft, Tulsa, OK) and Past software. The log-rank test (Statistica Software) was used to compare the Kaplan-Meier survival curves of each immune profile group of patients. Three distinct overall survivals were calculated from the following: i) date of the first cancer diagnosis of CRC for all patients (OS); ii) date of the first cancer diagnosis of CRC for nonmetastatic patients; and iii) date of the diagnosis of metastatic disease for metastatic patients (OS-MetDis) until death attributable to any cause. Because of the issue of multiplicity (three hypotheses), the Bonferroni correction was used and the significance threshold was set to be 0.0167. In a subset of 39 patients, immunohistochemical analysis of CD3 and CD8 expression was performed on FFPE tumor sections by using rabbit monoclonal CONFIRM anti-CD3 ready-to-use antibody clone 2GV6 or rabbit monoclonal CONFIRM anti-CD8 antibody clone SP57; sections were stained using the BenchMark ULTRA IHC/ISH System (all from Roche-Ventana Medical Systems, Tucson, AZ). The different marker expression in lymphocytes and in tumoral cells was evaluated independently by two investigators (C.U. and G.F.) on 10 high-power fields (×40 magnification) of tumoral core (CT) and on 10 high-power fields of infiltrative tumoral margins (IMs). The mean value of the different fields was then calculated. The overall CD3+ and CD8+ lymphocyte density (LyD) within the CT and the IM tumor compartments was calculated as follows: the mean of the four percentiles obtained for each marker (CD3 and CD8 mean counts at either CT or IM) was translated into three groups (low, intermediate, and high) corresponding to mean percentiles of 0% to <25%, ≥25% to <70%, and ≥70% to 100%, respectively.21Pagès F. Mlecnik B. Marliot F. Bindea G. Ou F.-S. Bifulco C. et al.International validation of the consensus immunoscore for the classification of colon cancer: a prognostic and accuracy study.Lancet. 2018; 391: 2128-2139Abstract Full Text Full Text PDF PubMed Scopus (1082) Google Scholar Immunohistochemical analysis of PD-1 and PD-L1 expression was performed on FFPE tumor sections using PD-1 (clone NAT105) mouse monoclonal antibody (Roche-Ventana Medical Systems) and PD-L1 (clone S263) Assay (rabbit monoclonal primary antibody) ready-to-use (Roche-Ventana Medical Systems). Sections were stained using the BenchMark ULTRA IHC/ISH System. The PD-1 and PD-L1 expression was evaluated in both lymphocytes and tumoral cells independently by two investigators (C.U. and G.F.) by considering one stained section entirely for each sample. PD-1 and PD-L1 expression in both tumor cells and lymphocytes was annotated as the estimated percentage of stained cells and was categorized as follows: negative/low (<1% stained tumor cells) and high (1% to 100% stained tumor cells).22Berntsson J. Eberhard J. Nodin B. Leandersson K. Larsson A.H. Jirström K. Expression of programmed cell death protein 1 (PD-1) and its ligand PD-L1 in colorectal cancer: relationship with sidedness and prognosis.Oncoimmunology. 2018; 7: e1465165Crossref PubMed Scopus (51) Google Scholar The clinical and pathologic characteristics of the samples included in the present study are the following: clinicopathologic stage at the time of diagnosis, primary tumor sidedness, presence of metastases, and, when available, presence of RAS or BRAF mutations. Seven dMMR CRCs were stage I, 45 were stage II, 15 were stage III, and 22 were stage IV at the time of diagnosis. Sixty-seven dMMR CRCs were right sided, and 22 were left sided; right-sided tumors were defined as arising from the cecum to the transverse colon, and left-sided tumors were defined as arising from the splenic flexure to the rectum. dMMR CRCs were divided into synchronous, metachronous, and nonmetastatic disease, with synchronous metastatic disease defined as distant metastasis occurring within 3 months since the primary diagnosis, metachronous metastatic disease defined as distant metastasis occurring beyond 3 months since the primary diagnosis, and nonmetastatic disease defined for the patients without metastases with at least 3 years of follow-up.23Mekenkamp L.J.M. Koopman M. Teerenstra S. van Krieken J.H.J.M. Mol L. Nagtegaal I.D. Punt C.J.A. Clinicopathological features and outcome in advanced colorectal cancer patients with synchronous vs metachronous metastases.Br J Cancer. 2010; 103: 159-164Crossref PubMed Scopus (110) Google Scholar In detail, 22 MMR CRCs were synchronous metastatic disease, 14 were metachronous metastatic disease, and 53 were nonmetastatic disease. Mutational status was known for 81 samples: 38 samples (46.9%) harbored a mutation within the BRAF gene (exon 15), 25 samples (30.9%) harbored a mutation within K-RAS or N-RAS genes (exons 12, 13, 61, 117, and 146), whereas 18 samples (22.2%) resulted to be wild type for both genes. The housekeeping genes selected for the normalization of the experiment presented a steady expression level in all the studied samples (data not shown). None of the samples was excluded after data normalization. Supplemental Figure S1 shows the heat maps of the mRNA normalized data, scaled to give all genes equal variance, generated via unsupervised clustering. Orange indicates high expression; blue indicates low expression. This plot is meant to provide a high-level exploratory view of the data. In detail, Supplemental Figure S1 shows the heat map displaying each sample level of the mRNA expression of all genes included in the Immune Profile Panel without any evidence of an overall mRNA expression clustering. Supplemental Figure S2 shows the heat maps displaying each sample level of expression of antigen processing, cytotoxicity, natural killer cell function, and T-cell functional groups of genes. All the 206 genes included in the four immune categories were then analyzed and classified by gene-expression functional clustering. Determination of the factors inside the immune profile panel analyzed plays a critical role in the elucidation of the linkage and mechanism between the immune system and the tumor samples analyzed. Therefore, a gene expression functional analysis was performed, generating four main factors denominated by ordinal number: 1, 2, 3, and 4 (Table 1). Factor 1 contains five genes constituting the antigen processing and presentation class II. Factor 2, composed by 13 entities, contains genes preferentially expressed in cytotoxic cells as surface markers in T cells, NK cells, and δγ T cells and gene-expressing inducer of cytolysis. Factor 3 is characterized by six genes involved in T-cell chemotaxis/cell adhesion. Finally, factor 4 contains four genes expressed in T-CD4+ regulator cells (Tregs). In summary, the results obtained from functional clustering analysis showed the antigen processing and presentation mechanism, CD4+ (Treg) population, cytotoxic activity, and cell adhesion as main functions of the immunity system in the samples analyzed.Table 1Functional Clustering Results, Descriptive Statistics, and Quartile Determination of the Analyzed Gene SetsFactormRNAClustering of the setsDescriptive statisticQuartile of gene expressionGene descriptionFDRP valueAcronymMean∗Log2 normalized counts.Median∗Log2 normalized counts.Range∗Log2 normalized counts.Quartile 2∗Log2 normalized counts.Quartile 1∗Log2 normalized counts.Quartile 3∗Log2 normalized counts.1. Antigen processing and presentationCD74CD74 molecule1.18×10–5MHC-II13.3313.4310.50–15.0713.4312.4714.05HLA-DMADM α1.18×10–59.929.837.25–12.259.869.0210.78HLA-DPA1DP α 11.18×10–511.1211.238.47–13.0811.1510.0412.13HLA-DPB1DP β 11.18×10–510.5210.548.25–12.3910.519.6611.58HLA-DRB3DR β 31.18×10–512.9112.8710.76–14.8412.7712.2413.702. NK/T-cell–mediated cytotoxicityCD8ACD8a molecule3.09×10–2Tc7.207.271.12–10.217.296.248.29PRF1Perforin 13.09×10–28.068.171.52–10.228.117.229.11GZMAGranzyme A3.09×10–28.438.461.52–11.128.417.279.62GZMBGranzyme B3.09×10–28.138.161.52–10.768.167.049.35GZMKGranzyme K3.09×10–26.896.971.52–9.557.025.917.96CD3DCD3d molecule3.09×10–27.907.985.14–9.877.937.018.67CD3ECD3e molecule3.09×10–27.237.431.12–9.367.416.2458.07CD27CD27 molecule3.09×10–27.127.271.52–9.557.206.0658.08CD2CD2 molecule3.09×10–25.495.591.12–7.835.534.946.53CD28CD28 molecule3.09×10–24.865.001.12–7.094.944.3355.64CD40LCD40 ligand3.09×10–25.095.081.52–7.517.075.1458.73ITGALIntegrin subunit α L3.09×10–27.817.905.24–9.657.836.8558.57CD80CD80 molecule3.09×10–25.665.771.52–8.075.715.196.413. T-cell chemotaxisCXCL9C-X-C motif chemokine ligand 91.62×10–5CAdh10.0210.196.28–12.9610.198.94511.31CXCL10C-X-C motif chemokine ligand 101.62×10–58.127.981.12–11.977.987.0559.66CXCL11C-X-C motif chemokine ligand 111.62×10–58.508.381.52–12.178.387.110.14KLRC1Killer cell lectin-like receptor C11.62×10–55.545.571.52–7.675.574.8556.21KLRC2Killer cell lectin-like receptor C21.62×10–56.376.491.12–8.846.495.667.27KLRD1Killer cell lectin-like receptor D11.62×10–55.465.601.52–7.695.604.9055.984. Regulation of CD4+ T cellsCD4CD4 molecule1×10–4Treg7.717.811.52–9.917.827.278.32CTLA-4Cytotoxic T-lymphocyte associated protein 41×10–47.207.291.52–9.107.186.458.33IL2RAIL-2 receptor subunit α1×10–45.275.051.12–8.137.716.848.51LAG3Lymphocyte activating 31×10–46.906.951.12–9.486.866.0258.13CAdh, cell adhesion; FDR, Benjamini-Yekutieli false discovery rate; MHC, major histocompatibility complex; NK, natural killer; Tc, cytotoxic T lymphocyte; Treg, T-regulatory.∗ Log2 normalized counts. Open table in a new tab CAdh, cell adhesion; FDR, Benjamini-Yekutieli false discovery rate; MHC, major histocompatibility complex; NK, natural killer; Tc, cytotoxic T lymphocyte; Treg, T-regulatory. To define the magnitude of activity of the four functional factors previously determined in relation to the analyzed samples, a quartile clustering technique has been used. For each of the factors and samples, a discrete value between −1 (low activity), 0 (medium activity), and 1 (high activity) has been assigned (Materials and Methods) based on the relative gene expression levels. Factor 1 [antigen processing and presentation, major histocompatibility complex (MHC)-II] showed down-regulation in 32 (36%) samples and up-regulation in 24 (27%). Factor 2, containing genes correlated with cytotoxic activity of the immune system, showed a slight difference between samples showing high activity (36; 40%) and low activity (32; 38%). Factor 3 (T-cell chemotaxis/cell adhesion) showed a large predominance of down-regulated genes (47; 51%) with respect to up-regulated samples (25; 28%). Finally, factor 4 (CD4+ Treg) displayed a uniform distribution between up-regulated (30; 32%) and down-regulated (28; 34%) samples. Tumor classification was performed using the results achieved from the quantitative quartile clustering of the samples described above. The algebraic sum of the discrete values of the singular factors was calculated, and a deregulation score (DS) with a value between −4 and 4 was determined for each sample. First, two groups of tumors samples were generated, on the basis of the DS values: COLD and NOT COLD. The COLD group contains tumor samples with a DS value ≤ −3 to sele
Referência(s)