A key molecular driver of tumor-infiltrating lymphocytes in invasive breast cancer identified by machine learning-based meta-mining

A key molecular driver of tumor-infiltrating lymphocytes in invasive breast cancer identified by machine learning-based meta-mining | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A key molecular driver of tumor-infiltrating lymphocytes in invasive breast cancer identified by machine learning-based meta-mining Chikako Honda, Sasagu Kurozumi, Graham R Ball, Ayaka Katayama, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7291997/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The immune system plays a crucial role at all stages of tumor development, including initiation, progression, and dissemination; however, the precise molecular mechanisms underlying tumor immunity remain unclear. In this study, we aimed to identify key targets associated with tumor-infiltrating lymphocytes (TILs) in early-stage breast cancer (BC) using a novel machine learning (ML) approach. We analyzed a cohort of 719 patients with early-stage BC from The Cancer Genome Atlas datasets, all of whom had available digital hematoxylin and eosin-stained whole slide images and tumor transcriptomic data. Stromal TIL grades (low, intermediate, and high) were evaluated based on the International Working Group criteria. Using artificial neural network ML methods, we identified 49 genes that exhibited differential expression across the stromal TIL grades. Cluster analysis of these genes resulted in the classification of patients into two distinct molecular subtypes (1 and 2), which were significantly associated with tumor aggressiveness and prognosis. Our findings highlight the potential of TIL-related gene sets in deciphering the intricate molecular networks that control tumor immunity in early-stage BC. tumor-infiltrating lymphocyte machine learning breast cancer prognostic marker Figures Figure 1 Figure 2 Figure 3 Introduction Breast cancer (BC) is the most common tumor in women, and early diagnosis and treatment are associated with a relatively favorable prognosis compared to other cancer types. However, some patients still experience poor outcomes. Therefore, several biomarkers capable of predicticting therapeutic response and prognosis have been investigated. Variations in immune responses within breast cancer tissues are believed to be influenced by the characteristics of tumor-infiltrating lymphocytes (TILs), which may impact drug sensitivity and prognosis [1].Previous retrospective studies suggest the potential use of TIL expression as a prognostic factor and a predictor of therapeutic efficacy in BC [1-3]. However, the complex molecular mechanisms of TILs have hindered the development of targeted therapies aimed at preventing metastatic recurrences. Researchers have attempted to decipher the complex gene networks associated with TILs using various statistical approaches [4, 5]; however, challenges remain, including the large number of target genes and high false discovery rates (FDRs), which impede validation analyses. To address these challenges, it is necessary to leverage big data sharing in BC and to develop and validate machine learning (ML) approaches for predicting biological processes related to TILs. Moreover, previous microarray and RNA sequencing analyses based on whole tumor samples may lack morphological assessment of tumor heterogeneity. Artificial neural networks (ANNs) play an important role in many ML algorithms, particularly for the analysis of high-dimensional biological data to extract key features related to clinical questions. One main advantage of ANNs is their ability to analyze datasets without being limited by linearity [6].They often yield generalizable predictions, exhibit low FDRs for biomarkers, and can identify biologically relevant markers. Several studies have demonstrated the utility of ANNs in identifying prognostic indicators and biomarkers for BC [7].In this study, we used a simplified architecture containing five hidden nodes and single transcript inputs to analyze high-dimensional transcriptomic datasets, which is a method that has been validated [8,9].This approach offers advantages over deep learning methods, including speed (screening up to 40 million models per hour), selectivity (maximizing marker separability to reduce false discovery risk), and the elimination of redundant nodes (up to 90% of nodes in deeper networks can be removed without loss of performance or increased false discovery risk) [10]. In this study, we mined gene sets associated with TIL expression using ANN systems with large transcriptomic datasets (The Cancer Genome Atlas; TCGA) in early-stage BC. In addition, we identified gene candidates that predict prognosis and therapeutic response based on these gene sets. Materials and Methods Clinicopathological data and mRNA data from TCGA cohort Data from The Cancer Genome Atlas (TCGA) cohort of female patients with BC ( n = 719) were extracted from the Genomic Data Commons Data Portal and the cBioPortal website [ 11 ]. The dataset included mRNA expression data evaluated by RNASeqV2, along with information on clinicopathological factors and prognosis. Digital hematoxylin and eosin (H&E)-stained slides for the TCGA_BRCA cohort were accessed from the cBioPortal website. Evaluation of TILs The percentage of stromal TILs (str-TILs) was assessed using light microscopy at 200–400× magnification on H&E-stained surgical specimens. Str-TILs were defined as mononuclear cells localized in the stromal tissue of BC. Based on the International TILs Working Group guidelines [ 12 ], str-TIL counts were classified into three levels: low (0–10%), intermediate (10–40%), and high (40–90%) [ 12 ]. Among the 719 patients with matched clinical, genetic data and H&E images, 151 were initially evaluated by CKH and then independently validated by AK, a breast pathologist. The correlation between the assessments of these two evaluators was measured using the kappa value, which showed significant consistency (kappa value = 0.402). For the remaining 568 patients, CKH assessed the str-TIL counts. Machine learning platform An ANN multilayer perceptron architecture with three layers was used. The input layer represented the expression of a single gene; the hidden layer consisted of five hidden nodes with a TanH activation function; and the output node, representing the TIL grade, also used a TanH activation function. During training, the weights of the architecture were updated using the Levenberg–Marquardt algorithm with regularization. Monte Carlo cross-validation (MCCV) was conducted using a 60:20:20 split, and early stopping was implemented during cross-validation. The training process was repeated for each transcript, creating a population of single transcript models ranked based on the performance of unseen test data. This process was iterated across 20 MCCVs, and concordance among the top 500 transcripts identified an enriched set associated with TIL grade. Artificial neural network inference Enriched genes associated with TILs were inputted into the ANN inference algorithm [ 13 ]. This approach leveraged the aforementioned algorithmic techniques to examine interactions between transcripts within a defined set, moving beyond gene lists to examine the interrelationships among clinical factors. By assessing the relative strength, direction (e.g., A influences B by an amount X and B influences A by an amount Y), sign (positive or negative influence), and relative magnitude of interactions between gene pairs, an interaction matrix was generated for the enriched set. Examination of this interaction matrix enabled the identification of the most influential and influenced features within the enriched set from a systems-level perspective. Bioinformatics analysis The TCGA cohort ( n = 719) was randomly divided into two groups, cohorts A and B, to ensure homogeneity in clinicopathological characteristics. The clinicopathological characteristics of cohorts A and B are summarized in Table S1. Each cohort was then further divided into 20 subgroups, with 20 patients in each subgroup, to identify differentially expressed genes (DEGs) among low, intermediate, and high TIL grade groups using an ANN approach. The top 500 genes with the strongest association with TIL expression were extracted using the ANN system following 20 iterations. The results were compiled separately for cohorts A and B (Table S2). Finally, overlapping genes in cohorts A and B were identified as the TIL-related gene set. The cluster 3.0 package was used for clustering and heatmap construction [ 14 ]. Cluster analysis was performed using the TCGA data, which were log2-transformed prior to analysis. For pathway analysis, Gene Set Enrichment Analysis and Molecular Signatures Database v7.4 were used to identify significantly enriched gene ontologies and pathways associated with the genes under investigation ( https://www.gsea-msigdb.org/gsea/index.jsp ). Transcriptomic and proteomic analysis of the inducible co-stimulator and its ligand The molecular taxonomy of the BC International Consortium (METABRIC) dataset [ 15 ], comprising 1,980 patients, was used for mRNA analysis of the inducible co-stimulator ( ICOS ) and ICOS ligand ( ICOSL ). The median mRNA value was used as the cut-off point for this analysis. Immunohistochemistry (IHC) was performed using a rabbit monoclonal anti-ICOS antibody (ab105227; Abcam) and a mouse monoclonal anti-ICOSL antibody (ab268059; Abcam) to assess the expression of ICOS and ICOSL. Deparaffinized sections were incubated in target retrieval solution, after which diluted primary antibodies (ICOS: 1:50, ICOSL: 1:10) were added and incubated overnight at 4°C. This was followed by incubation with a secondary antibody (Histofine Simple Stain MAX PO Multi) at room temperature for 30 min. First, IHC was conducted on full-faced slides from five patients with triple-negative BC (TNBC) to assess the expression of ICOS and ICOSL morphologically. IHC was also performed on tissue microarray (TMA) slides from a cohort of 41 TNBC cases to evaluate the clinicopathological characteristics of ICOS and ICOSL expression. In the TMA cohort, ICOS and ICOSL expression were evaluated in two TMA cores for each patient. Patients with > 1% cytoplasmic and/or membrane staining in stromal cells were classified as positive for stromal-ICOS and stromal-ICOSL in the TMA. The staining intensity for ICOSL expression in cancer cells was scored as follows: 0 (no staining or staining in 10% of tumor cells), 2 (moderate staining in > 10% of tumor cells), and 3 (strong staining in > 10% of tumor cells). Patients with an average expression value in two TMA scores above 1.5 were considered ICOSL-positive, whereas tumors with a score of 1.0 or less were considered ICOSL-negative. Statistical analysis Statistical analyses were conducted using IBM Statistical Package for the Social Sciences Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY, USA). A chi-square test was used to assess differences in various clinicopathological factors. Survival analysis, including univariate and multivariate analyses based on 10-year overall survival, was conducted using the Kaplan–Meier method with the log-rank test and the Cox proportional hazards regression model. Results Clinicopathological study of TILs expression In TCGA cohort of 719 patients, the distribution of TIL grades was 332 (46.2%), 224 (31.2%), and 163 (22.7%) for the low, intermediate, and high expression groups, respectively (Fig. 1 ). The TIL levels were significantly increased in estrogen receptor (ER)-negative ( p < 0.0001), progesterone receptor (PgR)-negative ( p < 0.0001), histological grade 3 ( p < 0.0001), high PD-L1 mRNA expression ( p < 0.0001), and TNBC subtype ( p < 0.0001; Table S3) specimens. Although TIL levels were not a significant prognostic factor across the cohort of 719 patients with BC, they were notably elevated in highly proliferative breast cancer subtypes, including TNBC and histological grade 3 types. Tumor-infiltrating lymphocytes-related gene set Table S2 lists the enriched DEGs associated with TILs identified through ML using ANNs. The analysis revealed 49 overlapping DEGs between cohorts A ( n = 400) and B ( n = 319) (Table 1 ). These 49 genes were significantly linked to gene ontologies related to immune system processes, including GO:005615 (extracellular space), GO:0072562 (blood microparticles), and GO:0031012 (extracellular matrix). In addition, Table S4 presents the top 10 molecular pathways identified among the significant pathways based on the FDR. Table 1 The 49 overlapping differentially expressed genes associated with tumor-infiltrating lymphocytes between cohorts A ( n = 400) and B ( n = 319) Entrez Gene ID Gene Symbol Gene Description 9447 AIM2 Absent in melanoma 2 64919 BCL11B BAF chromatin remodeling complex subunit BCL11B 51411 BIN2 Bridging integrator 2 6352 CCL5 C–C motif chemokine ligand 5 729230 CCR2 C–C motif chemokine receptor 2 914 CD2 CD2 molecule 919 CD247 CD247 molecule 952 CD38 CD38 molecule 10225 CD96 CD96 molecule 1493 CTLA4 Cytotoxic T-lymphocyte-associated protein 4 1520 CTSS Cathepsin S 10663 CXCR6 C–X–C motif chemokine receptor 6 115352 FCRL3 Fc receptor like 3 50943 FOXP3 Forkhead box P3 2633 GBP1 Guanylate-binding protein 1 3002 GZMB Granzyme B 29851 ICOS Inducible T-cell costimulator 3594 IL12RB1 Interleukin 12 receptor subunit beta 1 8807 IL18RAP Interleukin 18 receptor accessory protein 3561 IL2RG Interleukin 2 receptor subunit gamma 203522 INTS6L Integrator complex subunit 6 like 3662 IRF4 Interferon regulatory factor 4 3702 ITK IL2 inducible T-cell kinase 27074 LAMP3 Lysosomal-associated membrane protein 3 54900 LAX1 Lymphocyte transmembrane adapter 1 4049 LTA Lymphotoxin alpha 197259 MLKL Mixed lineage kinase domain-like pseudokinase 10943 MSL3 MSL complex subunit 3 84166 NLRC5 NLR family CARD domain containing 5 80380 PDCD1LG2 Programmed cell death 1 ligand 2 11040 PIM2 Pim-2 proto-oncogene, serine/threonine kinase 5450 POU2AF1 POU class 2 homeobox associating factor 1 5579 PRKCB Protein kinase C beta 5698 PSMB9 Proteasome 20S subunit beta 9 26191 PTPN22 Protein tyrosine phosphatase nonreceptor type 22 5778 PTPN7 Protein tyrosine phosphatase nonreceptor type 7 23231 SEL1L3 SEL1L family member 3 10507 SEMA4D Semaphorin 4D 5272 SERPINB9 Serpin family B member 9 4068 SH2D1A SH2 domain containing 1A 55423 SIRPG Signal-regulatory protein gamma 84174 SLA2 Src-like adapter 2 56833 SLAMF8 SLAM family member 8 124460 SNX20 Sorting nexin 20 11262 SP140 SP140 nuclear body protein 201633 TIGIT T-cell immunoreceptor with Ig and ITIM domains 3604 TNFRSF9 TNF receptor superfamily member 9 6846 XCL2 X-C motif chemokine ligand 2 79413 ZBED2 Zinc finger BED-type containing 2 Hierarchical clustering was used to analyze the 49 genes based on their expression similarities (Fig. 2 a). The cluster analysis classified the 719 patients into two subtypes: Subtype 1 ( n = 256, 35.6%) and subtype 2 ( n = 463, 64.4%). Subtype 2 exhibited a significantly higher TIL grade than subtype 1 ( p < 0.0001; Table 2 ). Table 2 Clinicopathological characteristics of subtypes 1 and 2 Factors TIL-related genomic subtype Total Significance Subtype 1 Subtype 2 P -value TIL grade High 27 (16.6%) 136 (83.4%) 163 < 0.0001 Intermediate 70 (31.3%) 154 (68.8%) 224 Low 159 (47.9%) 173 (52.1%) 332 PD-L1 ( CD274 ) mRNA High 67 (18.3%) 300 (81.7%) 367 < 0.0001 Low 189 (53.7%) 163 (46.3%) 352 ER Positive 209 (38.4%) 335 (61.6%) 5444 0.00024 Negative 34 (22.5%) 117 (77.5%) 151 PgR Positive 185 (39.7%) 281 (60.3%) 466 0.00024 Negative 57 (25.4%) 167 (74.6%) 224 HER2 Positive 34 (30.4%) 78 (69.6%) 112 0.27 Negative 174 (36.3%) 305 (63.7%) 479 Subtype HR-positive and HER2-negative 156 (40.4%) 230 (59.6%) 386 0.00034 HER2-positive 34 (30.4%) 78 (69.6%) 112 Triple negative 18 (19.4%) 75 (80.6%) 93 Tumor size pT2-4 184 (35.5%) 335 (64.5%) 519 0.93 pT1 72 (36.0%) 128 (64.0%) 200 Nodal status Positive 116 (33.2%) 233 (66.8%) 349 0.27 Negative 136 (37.3%) 229 (62.7%) 365 Histological grade Grade 3 75 (25.3%) 222 (74.7%) 297 < 0.0001 Grade 1, 2 171 (43.7%) 220 (56.3%) 391 Abbreviations: HR, hormonal receptor; ER, estrogen receptor; PgR, progesterone receptor; HER2, human epidermal growth factor 2; TIL, tumor-infiltrating lymphocyte Clinicopathological and prognostic significance of tumor-infiltrating lymphocyte-related gene sets In the cohort of 719 TCGA samples, subtype 2 was significantly associated with histological grade 3 ( p < 0.0001), ER-negative ( p = 0.00024), PgR-negative ( p = 0.00024), and high PD-L1 mRNA expression ( p < 0.0001) groups (Table 2 ). Interestingly, 80.6% of TNBCs were classified as subtype 2. In contrast, patients with TIL-related subtype 2 showed a significantly better prognosis than those with subtype 1 tumors (hazard ratio: 1.86, 95% confidence interval [CI]: 1.18–2.93, p = 0.0075; Fig. 2 b). In a multivariate survival analysis, the TIL-associated genomic subtype emerged as a strong, independent adverse prognostic factor (hazard ratio: 2.27, 95% CI: 1.11–4.64, p = 0.025; Table S5). Significance of ICOS in tumor-infiltrating lymphocyte-related gene sets The 49 genes in the TIL-related gene set were strongly associated with ICOS. The top interactions among the enriched TIL-related gene set identified several highly influential hubs (Fig. 3 a), including ICOS , SLA2 , ZBED2 , CD96 , CD38 , GZMB , and XCL1 . Although filtered into the top interactions, these hubs were connected to numerous other interactions. The unfiltered interaction matrix was further analyzed by mathematically collapsing it into total values, leading to each gene. The genes were ranked based on total influence (Table S6), with ICOS identified as the most influential gene among the enriched genes. Clinicopathological and prognostic significance of ICOS and its ligand at the mRNA level In the METABRIC cohort, high ICOS mRNA expression was significantly associated with high ICOSL mRNA expression ( p < 0.0001), ER-negativity ( p < 0.0001), PgR-negativity ( p < 0.0001), human epidermal growth factor receptor 2 (HER2)-positivity ( p < 0.0001), node-positivity ( p = 0.0052), and high histological grade ( p < 0.0001; Table S7). Patients with low ICOS mRNA expression in TNBC had a worse prognosis than those with high ICOS mRNA expression (Fig. 3 b). Survival analysis of ICOS expression in other BC subtypes is shown in Figure S1. In contrast, ICOSL mRNA expression was not a significant prognostic factor for TNBC (Fig. S2). Immunohistochemical expression of the ICOS and its ligand protein in triple-negative BC ICOS protein expression in cancer cells was uniformly weak, whereas positive ICOS expression was observed in stromal cells, including lymphocytes surrounding the cancer cells. ICOSL protein expression was positive in the cytoplasm of cancer cells and uniformly positive in stromal cells (Fig. 3 c, d). In a cohort of 41 patients with TNBC, 27 (65.9%) were positive and 14 (34.1%) were negative for ICOS. ICOS positivity was significantly associated with high TIL ( p = 0.020) and PD-L1 positivity ( p = 0.046; Table 3 ). ICOSL expression in stromal cells was positive in 95.1% of the 41 patients with TNBC. In contrast, ICOSL expression in cancer cells was positive in 51.2%, but did not correlate with ICOS positivity in stromal cells (Table 3 ). Table 3 Relationship between inducible co-stimulator expression and clinicopathological features. Factors ICOS (Stromal Cells) Total Significance Positive Negative P -value ICOSL (cancer cells) Positive 11 (52.4%) 10 (47.6%) 21 0.10 Negative 16 (80.0%) 4 (20.0%) 20 ICOSL (stromal cells) Positive 26 (66.7%) 13 (33.3%) 39 1.00 Negative 1 (50.0%) 1 (50.0%) 2 TIL grade High and intermediate 17 (85.0%) 3 (15.0%) 20 0.020 Low 10 (47.6%) 11 (52.4%) 21 PD-L1 (immune cells) Positive ( ≥ 1%) 16 (84.2%) 3 (15.8%) 19 0.046 Negative (< 1%) 11 (50.0%) 11 (50.0%) 22 Tumor size pT2-4 14 (60.9%) 9 (39.1%) 23 0.52 pT1 13 (72.2%) 5 (27.8%) 18 Nodal status Positive 8 (80.0%) 2 (20.0%) 10 0.45 Negative 19 (61.3%) 12 (38.7%) 31 Abbreviations: ICOS, inducible co-stimulator; ICOSL, inducible co-stimulator ligand; TIL, tumor-infiltrating lymphocyte In the 41 patients with TNBC, ICOS-positive patients tended to have better overall survival compared with ICOS-negative patients; however, this difference was not significant (Fig. 3 e). Conversely, ICOSL positivity in cancer cells was a significantly poor prognostic factor based on a univariate analysis ( p = 0.040; Fig. 3 f). Survival curves stratified by ICOS levels on stromal cells and ICOSL levels in cancer cells are shown in Figure S3. Discussion PD-1 is an immune checkpoint molecule containing an immunoglobulin-like domain. It is expressed on the surface of activated lymphocytes (T and B cells) and myeloid cells [ 16 , 17 ]. It binds to the PD-L1 ligand and inhibits lymphocyte activation. PD-1 is expressed in various cancer tissues and plays a major role in the immune evasion mechanisms of cancer cells [ 18 , 19 ]. Nivolumab and pembrolizumab are fully human monoclonal antibodies that target PD-1, whereas atezolizumab is a humanized monoclonal antibody that targets PD-L1 [ 20 – 22 ]. PD-1 and PD-L1 inhibitors have demonstrated notable antitumor effects in various cancers, including lung, gastrointestinal, melanoma, and BC, and are used as effective cancer treatments in clinical practice [ 23 – 25 ]. Some immune checkpoint inhibitors exhibit high efficacy and long-term survival after treatment discontinuation, whereas others show no efficacy. Therefore, identifying biomarkers that can predict efficacy is important. At present, PD-L1 IHC expression is used as a companion diagnostic biomarker to predict the efficacy of immune checkpoint inhibitors [ 26 ]; however, guidelines for the type of PD-L1 antibody and evaluation method are lacking. In addition, assessing PD-L1 expression alone may not be sufficient to evaluate the complex tumor immune mechanisms. In contrast, TILs are thought to play an important role in tumor immunity mechanisms in cancer tissues [ 3 ]. Several retrospective studies have demonstrated that TIL status is valuable in predicting prognosis and response to drug therapy in cancer, making the assessment of TILs increasingly important in clinical practice [ 1 , 2 , 27 ]. TILs provide information about dynamic tumor immune mechanisms and have been implicated in predicting responses to immune checkpoint inhibitors [ 28 , 29 ]. TILs compromise of various immune cell types and exhibit distinct characteristic patterns associated with prognosis and prediction of response to drug therapy [ 30 ]. Therefore, numerous molecular biology studies have attempted to elucidate the molecular biological characteristics related to TILs [ 31 ]. However, due to the complexity of the TIL pathway, additional key discoveries remain elusive. In the present study, 49 genes were found to be associated with TIL expression, showing significant correlation with TIL and PD-L1 expression. The discovery of this gene set linked to TILs may result in the identification of novel predictors of immune checkpoint inhibitor efficacy through further functional analyses. Furthermore, it may provide insights into the mechanisms of TIL induction and the intricate landscape of tumor immunity. Due to the intricate molecular pathways involved in immune responses within cancer tissues, elucidating these complex mechanisms manually remains challenging. Graham et al . used artificial neural network algorithms to identify candidate proliferation-related genes and evaluated their associations with clinicopathological features and outcomes in patients with BC [ 7 ]. This approach allows for the simultaneous analysis of a vast array of multimodal data, including genetic information (such as DNA copy number abnormalities, RNA transcription, and protein expression), clinicopathological characteristics, and prognostic data for ML to identify predictive markers for treatment response and prognosis in BC [ 32 ]. The present study integrated morphological characteristics of BC, particularly TILs, with extensive genetic information, leveraging a digital platform with ML capabilities. Digital technology was used to unravel the intricate molecular pathways controlling cancer immunity; however, challenges persist, including variability in optimal computational models and the identification of underlying biological mechanisms. Although digital technologies have been used in developing multigene profiling tools to guide chemotherapeutic treatment [ 33 , 34 ], tools targeting tumor immunity have not yet proven clinically useful. The TIL-related gene expression patterns identified in this study enabled the classification of invasive BC into two distinct subgroups, revealing a significant difference in prognosis between these groups. This finding holds promise for the development of novel gene signatures targeting tumor immunity. In this study, the TIL-associated gene set was functionally linked to ICOS, a member of the CD28 family, such as PD-1 and cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4). However, ICOS is distinct as it is expressed on activated T cells [ 35 ]. ICOS interacts with its specific ligand, ICOSL, which plays an important role in the differentiation of memory and effector T cells and immune responses [ 36 ]. The involvement of the ICOS/ICOSL pathway in tumor growth depends on the cytokines secreted. Activation of ICOS/ICOSL leads to an increase in CD4 + ICOS + and CD8 + ICOS + T cells, which increases the effector T-cell to Treg ratio [ 37 , 38 ]. Conversely, ICOS/ICOSL signaling promotes Treg differentiation from CD4 + T cells, which enhances the immunosuppressive effects [ 39 ]. High ICOS expression in T cells is a significant biomarker for predicting clinical responses in patients with cancer treated with anti-CTLA-4 or anti-PD-1 antibodies. Mice treated with anti-ICOS antibodies showed a weak response to anti-CTLA-4 antibodies, suggesting a complex interplay between CTLA-4, PD-1, and ICOS costimulatory molecules in immune checkpoint inhibitor therapy [ 40 ]. Analysis of the METEBRIC cohort revealed significantly higher ICOS mRNA expression in BCs with high proliferative potential, such as high histological grade, hormone receptor negativity, and a basal subtype. In the present study, elevated ICOS expression was associated with a favorable prognosis in TNBC. In contrast, an increase in ICOS + Tregs in TILs has been linked to a poor prognosis in gastric cancer [ 41 ]. The expression profile of ICOS/ICOSL in cancer tissues may dictate the mechanisms of tumor growth and antitumor immunity. Further studies are needed to elucidate the role of ICOS in antitumor immune responses. Conclusion The mechanism that induces TIL expression remains unclear, as TILs are a cell population with multifaceted functions. This study used ANNs to identify a set of genes associated with TILs through comprehensive RNA analysis of early-stage invasive BC. For effective therapeutic efficacy, immune checkpoint inhibitors require the presence of TILs surrounding tumors. This study aims to develop a gene panel that can facilitate the precise classification of patients who respond well to immune checkpoint inhibitors. Future studies are required to validate the utility of our TIL-related gene set in patient cohorts treated with immune checkpoint inhibitors. Furthermore, our results indicated that ICOS was the most affected gene in the TIL-related gene set. Thus, the significance of ICOS in TNBC warrants further study. Understanding the intricate molecular pathways involved in cancer progression through advanced digital technologies, as demonstrated in this study, may lead to the identification of novel therapeutic targets for cancer treatment [ 42 ]. Declarations Acknowledgments: The authors gratefully acknowledge the work of our research technician, Kumiko Sudo. Author Contributions: Conceptualization, C.K.H., S.K., and G.R.B.; Methodology, S.K., T.F., and G.R.B.; Investigation, C.K.H., A.K., and S.K.; Original Draft Preparation, C.K.H. and S.K.; Supervision, S.O., T.Y., T.O., J.H., K.S., and T.F. Funding: This research was funded by SPS KAKENHI (20K16375). Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Gunma University (HS2020-195). Informed Consent Statement: Patient consent was waived for studies using public databases. Conflict of Interest: SK has received honoraria from Kyowa Kirin Co., Ltd, Daiichi Sankyo Co., Ltd, Taiho Pharmaceutical Co., Ltd, Eli Lilly and Company, MSD K.K., AstraZeneca K.K., Chugai Pharmaceutical, Ltd., Dinow, Inc., Eisai Co., Ltd., Takeda Pharmaceutical Co., Ltd., and Pfizer Japan Inc. No conflicts of interest have been reported for any of the other authors. GB is the chief scientific officer of Intelligent Omics Ltd, which uses the algorithms presented here to identify biological drug targets. Data Availability Statement: Publicly available datasets were analyzed in this study. These data can be found here: https//www.cbioportal.org. Disclaimer/Publisher’s Note: The statements, opinions, and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions, or products referred to in the content. References Denkert C, von Minckwitz G, Brase JC, Sinn BV, Gade S, Kronenwett R, Pfitzner BM, Salat C, Loi S, Schmitt WD, Schem C, Fisch K, Darb-Esfahani S, Mehta K, Sotiriou C, Wienert S, Klare P, André F, Klauschen F, Blohmer J-U, Krappmann K, Schmidt M, Tesch H, Kümmel S, Sinn P, Jackisch C, Dietel M, Reimer T, Untch M, Loibl S (2015) Tumor-infiltrating lymphocytes and response to neoadjuvant chemotherapy with or without carboplatin in human epidermal growth factor receptor 2–positive and triple-negative primary breast cancers. J Clin Oncol 33:983–991. https://doi.org/10.1200/jco.2014.58.1967 Kurozumi S, Fujii T, Matsumoto H, Inoue K, Kurosumi M, Horiguchi J, Kuwano H (2017) Significance of evaluating tumor-infiltrating lymphocytes (TILs) and programmed cell death-ligand 1 (PD-L1) expression in breast cancer. Med Mol Morphol 50:185–194. https://doi.org/10.1007/s00795-017-0170-y Denkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner BI, Weber KE, Budczies J, Huober J, Klauschen F, Furlanetto J, Schmitt WD, Blohmer J-U, Karn T, Pfitzner BM, Kümmel S, Engels K, Schneeweiss A, Hartmann A, Noske A, Fasching PA, Jackisch C, van Mackelenbergh M, Sinn P, Schem C, Hanusch C, Untch M, Loibl S (2018) Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol 19:40–50. https://doi.org/10.1016/s1470-2045(17)30904-x Chen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA (2018) profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol Biol 1711:243–259. https://doi.org/10.1007/978-1-4939-7493-1_12 Danaher P, Warren S, Dennis L, D’Amico L, White A, Disis ML, Geller MA, Odunsi K, Beechem J, Fling SP (2017) Gene expression markers of tumor infiltrating leukocytes. J Immunother Cancer 5:186. https://doi.org/10.1101/068940 Lancashire LJ, Lemetre C, Ball GR (2009) An introduction to artificial neural networks in bioinformatics–application to complex microarray and mass spectrometry datasets in cancer studies. Briefings Bioinf 10:315–329. https://doi.org/10.1093/bib/bbp012 Abdel-Fatah TMA, Agarwal D, Liu D-X, Russell R, Rueda OM, Liu K, Xu B, Moseley PM, Green AR, Pockley AG, Rees RC, Caldas C, Ellis IO, Ball GR, Chan SYT (2016) SPAG5 as a prognostic biomarker and chemotherapy sensitivity predictor in breast cancer: a retrospective, integrated genomic, transcriptomic, and protein analysis. Lancet Oncol 17:1004–1018. https://doi.org/10.1016/s1470-2045(16)00174-1 Mian S, Ball G, Hornbuckle J, Holding F, Carmichael J, Ellis I, Ali S, Li G, McArdle S, Creaser C, Rees R (2003) A prototype methodology combining surface-enhanced laser desorption/ionization protein chip technology and artificial neural network algorithms to predict the chemoresponsiveness of breast cancer cell lines exposed to paclitaxel and doxorubicin under in vitro conditions. Proteomics 3:1725–1737. https://doi.org/10.1002/pmic.200300526 Smith BP, Brier ME (1996) Statistical approach to neural network model building for gentamicin peak predictions. J Pharm Sci 85:65–69. https://doi.org/10.1021/js950271l Tan M, Deklerck R, Jansen B, Bister M, Cornelis J (2011) A novel computer-aided lung nodule detection system for CT images. Med Phys 38:5630–5645. https://doi.org/10.1118/1.3633941 Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, Cerami E, Sander C, Schultz N (2013) Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 6:11. https://doi.org/10.1126/scisignal.2004088 Salgado R, Denkert C, Demaria S, Sirtaine N, Klauschen F, Pruneri G, Wienert S, Van den Eynden G, Baehner FL, Penault-Llorca F, Perez EA, Thompson EA, Symmans WF, Richardson AL, Brock J, Criscitiello C, Bailey H, Ignatiadis M, Floris G, Sparano J, Kos Z, Nielsen T, Rimm DL, Allison KH, Reis-Filho JS, Loibl S, Sotiriou C, Viale G, Badve S, Adams S, Willard-Gallo K, Loi S (2014) The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs working group 2014. Ann Oncol 26:259–271. https://doi.org/10.1093/annonc/mdu450 Tong DL, Boocock DJ, Dhondalay R, Lemetre GK, Ball C GR (2014) Artificial Neural Network Inference (ANNI): A study on gene-gene interaction for biomarkers in childhood sarcomas. PLoS ONE 9:e102483. https://doi.org/10.1371/journal.pone.0102483 de Hoon MJL, Imoto S, Nolan J, Miyano S (2004) Open source clustering software. Bioinformatics 20:1453–1454. https://doi.org/10.1093/bioinformatics/bth078 Curtis C, Shah SP, Chin S-F, Turashvili G, Rueda OM, Dunning MJ, Speed D, Lynch AG, Samarajiwa S, Yuan Y, Gräf S, Ha G, Haffari G, Bashashati A, Russell R, McKinney S, Langerød A, Green A, Provenzano E, Wishart G, Pinder S, Watson P, Markowetz F, Murphy L, Ellis I, Purushotham A, Børresen-Dale A-L, Brenton JD, Tavaré S, Caldas C, Aparicio S (2012) The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486:346–352. https://doi.org/10.1038/nature10983 Li C, Lin L, Hou M, Chu P (2021) PD–L1/PD–1 blockade in breast cancer: the immunotherapy era (Review). Oncol Rep 45:5–12. https://doi.org/10.3892/or.2020.7831 Parry RV, Chemnitz JM, Frauwirth KA, Lanfranco AR, Braunstein I, Kobayashi SV, Linsley PS, Thompson CB, Riley JL (2005) CTLA-4 and PD-1 receptors inhibit T-cell activation by distinct mechanisms. Mol Cell Biol 25:9543–9553. https://doi.org/10.1128/mcb.25.21.9543-9553.2005 Schütz F, Stefanovic S, Mayer L, von Au A, Domschke C, Sohn C (2017) PD-1/PD-L1 pathway in breast cancer. Oncol Res Treat 40:294–297. https://doi.org/10.1159/000464353 Keir ME, Butte MJ, Freeman GJ, Sharpe AH (2008) PD-1 and its ligands in tolerance and immunity. Annu Rev Immunol 26:677–704. https://doi.org/10.1146/annurev.immunol.26.021607.090331 Brahmer JR, Tykodi SS, Chow LQM, Hwu W-J, Topalian SL, Hwu P, Drake CG, Camacho LH, Kauh J, Odunsi K, Pitot HC, Hamid O, Bhatia S, Martins R, Eaton K, Chen S, Salay TM, Alaparthy S, Grosso JF, Korman AJ, Parker SM, Agrawal S, Goldberg SM, Pardoll DM, Gupta A, Wigginton JM (2012) Safety and activity of anti–PD-L1 antibody in patients with advanced cancer. N Engl J Med 366:2455–2465. https://doi.org/10.1056/nejmoa1200694 Adams S, Schmid P, Rugo HS, Winer EP, Loirat D, Awada A, Cescon DW, Iwata H, Campone M, Nanda R, Hui R, Curigliano G, Toppmeyer D, O’Shaughnessy J, Loi S, Paluch-Shimon S, Tan AR, Card D, Zhao J, Karantza V, Cortés J (2019) Pembrolizumab monotherapy for previously treated metastatic triple-negative breast cancer: cohort A of the phase II KEYNOTE-086 study. Ann Oncol 30:397–404. https://doi.org/10.1093/annonc/mdy517 Emens LA, Molinero L, Loi S, Rugo HS, Schneeweiss A, Diéras V, Iwata H, Barrios CH, Nechaeva M, Nguyen-Duc A, Chui SY, Husain A, Winer EP, Adams S, Schmid P (2021) Atezolizumab and nab -paclitaxel in advanced triple-negative breast cancer: biomarker evaluation of the IMpassion130 study. J Natl Cancer Instit 113:1005–1016. https://doi.org/10.1093/jnci/djab004 Ilson DH (2021) Adjuvant nivolumab in esophageal cancer — a new standard of care. N Engl J Med 384:1269–1271. https://doi.org/10.1056/nejme2101983 Forde PM, Spicer J, Lu S, Provencio M, Mitsudomi T, Awad MM, Felip E, Broderick SR, Brahmer JR, Swanson SJ, Kerr K, Wang C, Ciuleanu T-E, Saylors GB, Tanaka F, Ito H, Chen K-N, Liberman M, Vokes EE, Taube JM, Dorange C, Cai J, Fiore J, Jarkowski A, Balli D, Sausen M, Pandya D, Calvet CY, Girard N (2022) Neoadjuvant nivolumab plus chemotherapy in resectable lung cancer. N Engl J Med 386:1973–1985. https://doi.org/10.1056/nejmoa2202170 Bajorin DF, Witjes JA, Gschwend JE, Schenker M, Valderrama BP, Tomita Y, Bamias A, Lebret T, Shariat SF, Park SH, Ye D, Agerbaek M, Enting D, McDermott R, Gajate P, Peer A, Milowsky MI, Nosov A, Neif Antonio J, Tupikowski K, Toms L, Fischer BS, Qureshi A, Collette S, Unsal-Kacmaz K, Broughton E, Zardavas D, Koon HB, Galsky MD (2021) Adjuvant nivolumab versus placebo in muscle-invasive urothelial carcinoma. N Engl J Med 384:2102–2114. https://doi.org/10.1056/nejmoa2034442 Cortes J, Cescon DW, Rugo HS, Nowecki Z, Im S-A, Yusof MM, Gallardo C, Lipatov O, Barrios CH, Holgado E, Iwata H, Masuda N, Torregroza Otero M, Gokmen E, Loi S, Guo Z, Zhao J, Aktan G, Karantza V, Schmid P (2020) KEYNOTE-355 Investigators. Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355): a randomized, placebo-controlled, double-blind, phase 3 clinical trial. Lancet 396:1817–1828. https://doi.org/10.1200/jco.2020.38.15_suppl.1000 Honda C, Kurozumi S, Katayama A, Hanna–Khalil B, Masuda K, Nakazawa Y, Ogino M, Obayashi S, Yajima R, Makiguchi T, Oyama T, Horiguchi J, Shirabe K, Fujii T (2021) Prognostic value of tumor–infiltrating lymphocytes in estrogen receptor–positive and human epidermal growth factor receptor 2–negative breast cancer. Mol Clin Oncol 15:252. https://doi.org/10.3892/mco.2021.2414 El Bairi K, Haynes HR, Blackley E, Fineberg S, Shear J, Turner S, De Freitas JR, Sur D, Amendola LC, Gharib M, Kallala A, Arun I, Azmoudeh-Ardalan F, Fujimoto L, Sua LF, Liu SW, Lien HC, Kirtani P, Balancin M, El Attar H, Guleria P, Yang W, Shash E, Chen IC, Bautista V, Do Prado Moura JF, Rapoport BL, Castaneda C, Spengler E, Acosta-Haab G, Frahm I, Sanchez J, Castillo M, Bouchmaa N, Md Zin RR, Shui R, Onyuma T, Yang W, Husain Z, Willard-Gallo K, Coosemans A, Perez EA, Provenzano E, Ericsson PG, Richardet E, Mehrotra R, Sarancone S, Ehinger A, Rimm DL, Bartlett JMS, Viale G, Denkert C, Hida AI, Sotiriou C, Loibl S, Hewitt SM, Badve S, Symmans WF, Kim RS, Pruneri G, Goel S, Francis PA, Inurrigarro G, Yamaguchi R, Garcia-Rivello H, Horlings H, Afqir S, Salgado R, Adams S, Kok M, Dieci MV, Michiels S, Demaria S, Loi S (2021) International immuno-oncology biomarker working group. The tale of TILs in breast cancer: a. npj Breast Cancer 7:150 report from the International Immuno-Oncology Biomarker Working Group Hendry S, Salgado R, Gevaert T, Russell PA, John T, Thapa B, Christie M, van de Vijver K, Estrada MV, Gonzalez-Ericsson PI, Sanders M, Solomon B, Solinas C, Van den Eynden GGGM, Allory Y, Preusser M, Hainfellner J, Pruneri G, Vingiani A, Demaria S, Symmans F, Nuciforo P, Comerma L, Thompson EA, Lakhani S, Kim S-R, Schnitt S, Colpaert C, Sotiriou C, Scherer SJ, Ignatiadis M, Badve S, Pierce RH, Viale G, Sirtaine N, Penault-Llorca F, Sugie T, Fineberg S, Paik S, Srinivasan A, Richardson A, Wang Y, Chmielik E, Brock J, Johnson DB, Balko J, Wienert S, Bossuyt V, Michiels S, Ternes N, Burchardi N, Luen SJ, Savas P, Klauschen F, Watson PH, Nelson BH, Criscitiello C, O’Toole S, Larsimont D, de Wind R, Curigliano G, André F, Lacroix-Triki M, van de Vijver M, Rojo F, Floris G, Bedri S, Sparano J, Rimm D, Nielsen T, Kos Z, Hewitt S, Singh B, Farshid G, Loibl S, Allison KH, Tung N, Adams S, Willard-Gallo K, Horlings HM, Gandhi L, Moreira A, Hirsch F, Dieci MV, Urbanowicz M, Brcic I, Korski K, Gaire F, Koeppen H, Lo A, Giltnane J, Rebelatto MC, Steele KE, Zha J, Emancipator K, Juco JW, Denkert C, Reis-Filho J, Loi S, Fox SB (2017) Assessing tumor-infiltrating lymphocytes in solid tumors: a practical review for pathologists and proposal for a standardized method from the international immuno-oncology biomarkers working group: part 2: TILs in melanoma, gastrointestinal tract carcinomas, non–small cell lung carcinoma and mesothelioma, endometrial and ovarian carcinomas, squamous cell carcinoma of the head and neck, genitourinary carcinomas, and primary brain tumors. Adv Anat Pathol 24:311–335. https://doi.org/10.1097/pap.0000000000000161 Ahn S, Chung YR, Seo AN, Kim M, Woo JW, Park SY (2020) Changes and prognostic values of tumor-infiltrating lymphocyte subsets after primary systemic therapy in breast cancer. PLoS ONE 15:e0233037. https://doi.org/10.1371/journal.pone.0233037 Romagnoli G, Wiedermann M, Hübner F, Wenners A, Mathiak M, Röcken C, Maass N, Klapper W, Alkatout I (1936) Morphological evaluation of tumor-infiltrating lymphocytes (TILs) to investigate invasive breast cancer immunogenicity, reveal lymphocytic networks and help relapse prediction: a retrospective study. Int J Mol Sci 18:1936. https://doi.org/10.3390/ijms18091936 López-Cortés A, Cabrera-Andrade A, Vázquez-Naya JM, Pazos A, Gonzáles-Díaz H, Paz-y-Miño C, Guerrero S, Pérez-Castillo Y, Tejera E, Munteanu CR (2020) Prediction of breast cancer proteins involved in immunotherapy, metastasis, and RNA-binding using molecular descriptors and artificial neural networks. Sci Rep 10:8515. https://doi.org/10.1038/s41598-020-65584-y Woodhouse R, Li M, Hughes J, Delfosse D, Skoletsky J, Ma P, Meng W, Dewal N, Milbury C, Clark T, Donahue A, Stover D, Kennedy M, Dacpano-Komansky J, Burns C, Vietz C, Alexander B, Hegde P, Dennis L (2020) Clinical and analytical validation of FoundationOne liquid CDx, a novel 324-Gene cfDNA-based comprehensive genomic profiling assay for cancers of solid tumor origin. PLoS ONE 15:e0237802. https://doi.org/10.1371/journal.pone.0237802 Higami A, Takada M, Kawaguchi-Sakita N, Kawashima M, Kawaguchi K, Yamaguchi A, Takeuchi Y, Yamada Y, Toi M (2024) Predicting invasive disease-free survival in ER-positive, HER2-negative early breast cancer using the PAM50 risk-of-recurrence score: a retrospective analysis using single-center long-term follow-up data of postmenopausal Japanese patients. Int J Clin Oncol. 2024;29(11):1715–1720. https://doi.org/10.1007/s10147-024-02604-1 Hutloff A, Dittrich AM, Beier KC, Eljaschewitsch B, Kraft R, Anagnostopoulos I, Kroczek RA (1999) ICOS is an inducible T-cell co-stimulator structurally and functionally related to CD28. Nature 397:263–266. https://doi.org/10.1038/16717 Solinas C, Gu-Trantien C, Willard-Gallo K (2020) The rationale behind targeting the ICOS-ICOS ligand costimulatory pathway in cancer immunotherapy. ESMO Open 5:e000544. https://doi.org/10.1136/esmoopen-2019-000544 Marinelli O, Nabissi M, Morelli MB, Torquati L, Amantini C, Santoni G (2018) ICOS-L as a Potential Therapeutic Target for Cancer Immunotherapy. Curr Protein Pept Sc 19:1107–1113. https://doi.org/10.2174/1389203719666180608093913 Zhou W, Yu M, Pan H, Qiu W, Wang H, Qian M, Che N, Zhang K, Mao X, Li L, Wang R, Xie H, Ling L, Zhao Y, Liu X, Wang C, Ding Q, Wang S (2021) Microwave ablation induces Th1-type immune response with activation of ICOS pathway in early-stage breast cancer. J ImmunoTher Cancer 9:e002343. https://doi.org/10.1136/jitc-2021-002343 Sainson RCA, Thotakura AK, Kosmac M, Borhis G, Parveen N, Kimber R, Carvalho J, Henderson SJ, Pryke KL, Okell T, O'Leary S, Ball S, Van Krinks C, Gamand L, Taggart E, Pring EJ, Ali H, Craig H, Wong VWY, Liang Q, Rowlands RJ, Lecointre M, Campbell J, Kirby I, Melvin D, Germaschewski V, Oelmann E, Quaratino S, McCourt M (2020) An antibody targeting icos increases intratumoral cytotoxic to regulatory T-cell ratio and induces tumor regression. Cancer Immunol Res 8:1568–1582. https://doi.org/10.1158/2326-6066.cir-20-0034 Burlion A, Ramos RN, KC P, Sendeyo K, Corneau A, Ménétrier-Caux C, Piaggio E, Olive D, Caux C, Marodon G (2019) A novel combination of chemotherapy and immunotherapy controls tumor growth in mice with a human immune system. OncoImmunology 8:e1596005. https://doi.org/10.1080/2162402x.2019.1596005 Huang X, Liu X, Lin X, Yu H, Sun J, Liu X, Chen C, Jin H, Zhang G, Shi X, Zhang Q, Yu J (2014) Role of plasmacytoid dendritic cells and inducible costimulator-positive regulatory T cells in the immunosuppression microenvironment of gastric cancer. Cancer Sci 105:150–158. https://doi.org/10.1111/cas.12327 Kurozumi S, Ball GR (2024) Research on biomarkers using innovative artificial intelligence systems in breast cancer. Int J Clin Oncol 29:1669–1675. https://doi.org/10.1007/s10147-024-02602-3 Supplementary Materials Supplementary figures S1-S3 and supplementary tables S1-S7 are not available with this version. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7291997","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":496507114,"identity":"e887aa11-7973-413d-ba7d-8591346ba33b","order_by":0,"name":"Chikako Honda","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCklEQVRIie2RMUvEMBTHUwKX5dW5pfcVhHccdPHwvkok0CmDIrg4nC7XpR8gx30JXTpXAmbpLIoOd4uucZEbBI241KGxo2B+8Ah58OP9X0JIIPAXofRmw3EGUXnRbYNHYUuB9rgY06oZqkCbp8rqGVN8YK59VfAMkEK8ernLTuQhwYZuLRk/9Sr5vWimgCNI1/IsW9XCKaNpQuDZoxRcAAJM1rLI4pourhqSu7zao0jUgAnMH9ovZeGmsDe/0raTS4UIkWK3TtFOgV+mmKUgFjlElaQHcW1IquE04b5dNDU7/v4xj0qzfYzrc7Jnymv7WvW/WAfA75O6Ovr5sX2wTeeyG6QEAoHA/+AT89JUXeteHM0AAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0003-6918-1380","institution":"Takasaki General Medical Center","correspondingAuthor":true,"prefix":"","firstName":"Chikako","middleName":"","lastName":"Honda","suffix":""},{"id":496507115,"identity":"4f830b2d-8092-482e-9020-f5a5c7bc9522","order_by":1,"name":"Sasagu Kurozumi","email":"","orcid":"https://orcid.org/0000-0002-9971-2918","institution":"International University of Health and Welfare: Kokusai Iryo Fukushi Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Sasagu","middleName":"","lastName":"Kurozumi","suffix":""},{"id":496507116,"identity":"2c2024d0-7700-4ab9-8f86-7e3ec23f3104","order_by":2,"name":"Graham R Ball","email":"","orcid":"","institution":"intelligent Omics Ltd","correspondingAuthor":false,"prefix":"","firstName":"Graham","middleName":"R","lastName":"Ball","suffix":""},{"id":496507117,"identity":"6a170279-2ecf-40d0-8b24-67142dae2bb0","order_by":3,"name":"Ayaka Katayama","email":"","orcid":"","institution":"Shizuoka Cancer Center: Shizuoka Kenritsu Shizuoka Gan Center","correspondingAuthor":false,"prefix":"","firstName":"Ayaka","middleName":"","lastName":"Katayama","suffix":""},{"id":496507118,"identity":"6e37c4f6-27e2-4ba7-9e62-6c0a342c078b","order_by":4,"name":"Takehiko Yokobori","email":"","orcid":"","institution":"Gunma University: Gunma Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Takehiko","middleName":"","lastName":"Yokobori","suffix":""},{"id":496507119,"identity":"22fa316e-20ba-43e3-b4da-cb655b1364fe","order_by":5,"name":"Yukio Koibuchi","email":"","orcid":"","institution":"National Hospital Organization Takasaki General Medical Center Attached Takasaki School of Nursing: Kokuritsu Byoin Kiko Takasaki Sogo Iryo Center Fuzoku Takasaki Kango Gakko","correspondingAuthor":false,"prefix":"","firstName":"Yukio","middleName":"","lastName":"Koibuchi","suffix":""},{"id":496507120,"identity":"433eaf19-6088-4450-8098-406c5dc8865b","order_by":6,"name":"Tetsunari Oyama","email":"","orcid":"","institution":"Gunma University: Gunma Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Tetsunari","middleName":"","lastName":"Oyama","suffix":""},{"id":496507121,"identity":"c2c9c30d-8f42-4d70-b5f3-c9fb7e03d940","order_by":7,"name":"Jun Horiguchi","email":"","orcid":"","institution":"International University of Health and Welfare: Kokusai Iryo Fukushi Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Horiguchi","suffix":""},{"id":496507122,"identity":"3b448387-4d3a-47ae-89b4-f4a63d29b80e","order_by":8,"name":"Ken Shirabe","email":"","orcid":"","institution":"Gunma University: Gunma Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Ken","middleName":"","lastName":"Shirabe","suffix":""},{"id":496507123,"identity":"19165aa5-7f0d-43ab-ab21-de6e7f5ff5eb","order_by":9,"name":"Sayaka Obayashi","email":"","orcid":"","institution":"Gunma University: Gunma Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Sayaka","middleName":"","lastName":"Obayashi","suffix":""},{"id":496507124,"identity":"3a0e2c3a-b44e-41a7-a74f-d50692393b7a","order_by":10,"name":"Takaaki Fujii","email":"","orcid":"","institution":"Gunma University: Gunma Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Takaaki","middleName":"","lastName":"Fujii","suffix":""}],"badges":[],"createdAt":"2025-08-04 14:03:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7291997/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7291997/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88894486,"identity":"0f6d22e8-d07b-45b1-a091-b99119746de0","added_by":"auto","created_at":"2025-08-12 13:02:33","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":818526,"visible":true,"origin":"","legend":"\u003cp\u003eEvaluation of the grade of tumor-infiltrating lymphocytes using the International Tumor-Infiltrating Lymphocytes guidelines: (a) low grade, (b) intermediate grade, and (c) high grade\u003c/p\u003e","description":"","filename":"Figure1R1.png","url":"https://assets-eu.researchsquare.com/files/rs-7291997/v1/edffee38ac4a08cd7c7ca0e7.png"},{"id":88896984,"identity":"5f69e9b6-eccd-4643-96c8-0ebbb5c9aa28","added_by":"auto","created_at":"2025-08-12 13:10:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":932049,"visible":true,"origin":"","legend":"\u003cp\u003eCluster analysis of 49 tumor-infiltrating lymphocytes (TIL)-related genes for all cases\u003c/p\u003e","description":"","filename":"Figure2R1.png","url":"https://assets-eu.researchsquare.com/files/rs-7291997/v1/159e13de697159970a702118.png"},{"id":88894489,"identity":"f8af7e92-1909-426b-92b6-369e00065788","added_by":"auto","created_at":"2025-08-12 13:02:33","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2612061,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Cytoscape analysis of genes that affect tumor-infiltrating lymphocytes (TILs) and are affected by TILs; \u003cem\u003eICOS\u003c/em\u003e was the most associated with TIL grade among the TIL-related genes.\u003c/p\u003e\n\u003cp\u003e(b) In the METABRIC cohort, patients with high \u003cem\u003eICOS\u003c/em\u003e mRNA expression in the triple-negative type had a better prognosis compared with those with low ICOS mRNA expression.\u003c/p\u003e\n\u003cp\u003ec, d) The morphological characteristics of ICOS and ICOSL protein expression. (c) Immunostaining for ICOS shows predominant staining in stromal cells. d) Immunostaining of ICOSL reveals staining in both tumor and immune cells infiltrating the stroma.\u003c/p\u003e\n\u003cp\u003ee, f) Forty-one patients with triple-negative breast cancer who underwent surgery at our hospital showing (e) that the prognosis tended to be better in the group with high ICOS expression and (f) that the group with high ICOSL expression had a significantly worse prognosis.\u003c/p\u003e","description":"","filename":"Figure3R1.png","url":"https://assets-eu.researchsquare.com/files/rs-7291997/v1/29d3d491b63b4f8b16ef232b.png"},{"id":89582220,"identity":"767575df-1592-407d-81bc-d46976016bec","added_by":"auto","created_at":"2025-08-21 14:20:14","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6219708,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7291997/v1/ab5713fd-ec31-40a4-9035-3f827a40b5d6.pdf"}],"financialInterests":"","formattedTitle":"A key molecular driver of tumor-infiltrating lymphocytes in invasive breast cancer identified by machine learning-based meta-mining","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBreast cancer (BC) is the most common tumor in women, and early diagnosis and treatment are associated with a relatively favorable prognosis compared\u0026nbsp;to\u0026nbsp;other cancer types. However, some patients still experience poor outcomes. Therefore, several biomarkers capable of predicticting therapeutic response and prognosis have been\u0026nbsp;investigated.\u003c/p\u003e\n\u003cp\u003eVariations in immune responses within breast cancer tissues\u0026nbsp;are believed to\u0026nbsp;be influenced by the characteristics of tumor-infiltrating lymphocytes (TILs), which may\u0026nbsp;impact\u0026nbsp;drug sensitivity and prognosis [1].Previous retrospective studies\u0026nbsp;suggest\u0026nbsp;the potential\u0026nbsp;use\u0026nbsp;of TIL expression as a prognostic factor and a predictor of\u0026nbsp;therapeutic efficacy\u0026nbsp;in BC [1-3]. However, the complex molecular mechanisms\u0026nbsp;of\u0026nbsp;TILs have hindered the development of targeted therapies aimed at preventing\u0026nbsp;metastatic recurrences.\u0026nbsp;Researchers have attempted to\u0026nbsp;decipher\u0026nbsp;the complex gene networks associated with TILs using various statistical approaches [4, 5]; however, challenges remain, including the large number of target genes and high false discovery rates (FDRs), which impede validation\u0026nbsp;analyses. To address these challenges, it is necessary to leverage big data sharing in BC and to develop and validate machine learning (ML) approaches for predicting biological processes related to TILs. Moreover, previous microarray and RNA sequencing analyses based on whole tumor samples may lack morphological assessment of tumor heterogeneity.\u003c/p\u003e\n\u003cp\u003eArtificial neural networks (ANNs) play\u0026nbsp;an important\u0026nbsp;role in many ML algorithms, particularly\u0026nbsp;for the analysis of\u0026nbsp;high-dimensional biological data to extract key features related to clinical questions. One main\u0026nbsp;advantage\u0026nbsp;of ANNs is their ability to analyze datasets without being limited by linearity [6].They often yield generalizable predictions, exhibit low FDRs for biomarkers, and can identify biologically relevant markers. Several studies have demonstrated the utility of ANNs in identifying prognostic indicators and biomarkers for BC [7].In this study, we\u0026nbsp;used\u0026nbsp;a simplified architecture\u0026nbsp;containing\u0026nbsp;five hidden nodes and single transcript inputs to analyze high-dimensional transcriptomic datasets,\u0026nbsp;which is\u0026nbsp;a method that has been validated [8,9].This approach offers advantages over deep learning methods, including speed (screening up to 40 million models per hour), selectivity (maximizing marker separability to reduce false discovery risk), and\u0026nbsp;the\u0026nbsp;elimination of redundant nodes (up to 90% of nodes in deeper networks can be removed without loss of performance or increased false discovery risk) [10].\u003c/p\u003e\n\u003cp\u003eIn this study, we mined gene sets associated with TIL expression using ANN systems with large transcriptomic datasets (The Cancer Genome Atlas; TCGA) in early-stage BC. In addition, we identified gene candidates that predict prognosis and therapeutic response based on these gene sets.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cb\u003eClinicopathological data and mRNA data from TCGA cohort\u003c/b\u003e\u003c/p\u003e\u003cp\u003eData from The Cancer Genome Atlas (TCGA) cohort of female patients with BC (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;719) were extracted from the Genomic Data Commons Data Portal and the cBioPortal website [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. The dataset included mRNA expression data evaluated by RNASeqV2, along with information on clinicopathological factors and prognosis. Digital hematoxylin and eosin (H\u0026amp;E)-stained slides for the TCGA_BRCA cohort were accessed from the cBioPortal website.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEvaluation of TILs\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe percentage of stromal TILs (str-TILs) was assessed using light microscopy at 200\u0026ndash;400\u0026times; magnification on H\u0026amp;E-stained surgical specimens. Str-TILs were defined as mononuclear cells localized in the stromal tissue of BC. Based on the International TILs Working Group guidelines [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], str-TIL counts were classified into three levels: low (0\u0026ndash;10%), intermediate (10\u0026ndash;40%), and high (40\u0026ndash;90%) [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Among the 719 patients with matched clinical, genetic data and H\u0026amp;E images, 151 were initially evaluated by CKH and then independently validated by AK, a breast pathologist. The correlation between the assessments of these two evaluators was measured using the kappa value, which showed significant consistency (kappa value\u0026thinsp;=\u0026thinsp;0.402). For the remaining 568 patients, CKH assessed the str-TIL counts.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMachine learning platform\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAn ANN multilayer perceptron architecture with three layers was used. The input layer represented the expression of a single gene; the hidden layer consisted of five hidden nodes with a TanH activation function; and the output node, representing the TIL grade, also used a TanH activation function. During training, the weights of the architecture were updated using the Levenberg\u0026ndash;Marquardt algorithm with regularization. Monte Carlo cross-validation (MCCV) was conducted using a 60:20:20 split, and early stopping was implemented during cross-validation. The training process was repeated for each transcript, creating a population of single transcript models ranked based on the performance of unseen test data. This process was iterated across 20 MCCVs, and concordance among the top 500 transcripts identified an enriched set associated with TIL grade.\u003c/p\u003e\u003cp\u003e\u003cb\u003eArtificial neural network inference\u003c/b\u003e\u003c/p\u003e\u003cp\u003eEnriched genes associated with TILs were inputted into the ANN inference algorithm [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. This approach leveraged the aforementioned algorithmic techniques to examine interactions between transcripts within a defined set, moving beyond gene lists to examine the interrelationships among clinical factors. By assessing the relative strength, direction (e.g., A influences B by an amount X and B influences A by an amount Y), sign (positive or negative influence), and relative magnitude of interactions between gene pairs, an interaction matrix was generated for the enriched set. Examination of this interaction matrix enabled the identification of the most influential and influenced features within the enriched set from a systems-level perspective.\u003c/p\u003e\u003cp\u003e\u003cb\u003eBioinformatics analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe TCGA cohort (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;719) was randomly divided into two groups, cohorts A and B, to ensure homogeneity in clinicopathological characteristics. The clinicopathological characteristics of cohorts A and B are summarized in Table S1. Each cohort was then further divided into 20 subgroups, with 20 patients in each subgroup, to identify differentially expressed genes (DEGs) among low, intermediate, and high TIL grade groups using an ANN approach. The top 500 genes with the strongest association with TIL expression were extracted using the ANN system following 20 iterations. The results were compiled separately for cohorts A and B (Table S2). Finally, overlapping genes in cohorts A and B were identified as the TIL-related gene set.\u003c/p\u003e\u003cp\u003eThe cluster 3.0 package was used for clustering and heatmap construction [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Cluster analysis was performed using the TCGA data, which were log2-transformed prior to analysis.\u003c/p\u003e\u003cp\u003eFor pathway analysis, Gene Set Enrichment Analysis and Molecular Signatures Database v7.4 were used to identify significantly enriched gene ontologies and pathways associated with the genes under investigation (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/gsea/index.jsp\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/gsea/index.jsp\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cb\u003eTranscriptomic and proteomic analysis of the inducible co-stimulator and its ligand\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe molecular taxonomy of the BC International Consortium (METABRIC) dataset [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], comprising 1,980 patients, was used for mRNA analysis of the inducible co-stimulator (\u003cem\u003eICOS\u003c/em\u003e) and ICOS ligand (\u003cem\u003eICOSL\u003c/em\u003e). The median mRNA value was used as the cut-off point for this analysis.\u003c/p\u003e\u003cp\u003eImmunohistochemistry (IHC) was performed using a rabbit monoclonal anti-ICOS antibody (ab105227; Abcam) and a mouse monoclonal anti-ICOSL antibody (ab268059; Abcam) to assess the expression of ICOS and ICOSL. Deparaffinized sections were incubated in target retrieval solution, after which diluted primary antibodies (ICOS: 1:50, ICOSL: 1:10) were added and incubated overnight at 4\u0026deg;C. This was followed by incubation with a secondary antibody (Histofine Simple Stain MAX PO Multi) at room temperature for 30 min.\u003c/p\u003e\u003cp\u003eFirst, IHC was conducted on full-faced slides from five patients with triple-negative BC (TNBC) to assess the expression of ICOS and ICOSL morphologically. IHC was also performed on tissue microarray (TMA) slides from a cohort of 41 TNBC cases to evaluate the clinicopathological characteristics of ICOS and ICOSL expression. In the TMA cohort, ICOS and ICOSL expression were evaluated in two TMA cores for each patient. Patients with \u0026gt;\u0026thinsp;1% cytoplasmic and/or membrane staining in stromal cells were classified as positive for stromal-ICOS and stromal-ICOSL in the TMA. The staining intensity for ICOSL expression in cancer cells was scored as follows: 0 (no staining or staining in \u0026lt;\u0026thinsp;10% of tumor cells), 1 (weak staining in \u0026gt;\u0026thinsp;10% of tumor cells), 2 (moderate staining in \u0026gt;\u0026thinsp;10% of tumor cells), and 3 (strong staining in \u0026gt;\u0026thinsp;10% of tumor cells). Patients with an average expression value in two TMA scores above 1.5 were considered ICOSL-positive, whereas tumors with a score of 1.0 or less were considered ICOSL-negative.\u003c/p\u003e\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eStatistical analyses were conducted using IBM Statistical Package for the Social Sciences Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY, USA). A chi-square test was used to assess differences in various clinicopathological factors. Survival analysis, including univariate and multivariate analyses based on 10-year overall survival, was conducted using the Kaplan\u0026ndash;Meier method with the log-rank test and the Cox proportional hazards regression model.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eClinicopathological study of TILs expression\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn TCGA cohort of 719 patients, the distribution of TIL grades was 332 (46.2%), 224 (31.2%), and 163 (22.7%) for the low, intermediate, and high expression groups, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The TIL levels were significantly increased in estrogen receptor (ER)-negative (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), progesterone receptor (PgR)-negative (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), histological grade 3 (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), high \u003cem\u003ePD-L1\u003c/em\u003e mRNA expression (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), and TNBC subtype (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001; Table S3) specimens. Although TIL levels were not a significant prognostic factor across the cohort of 719 patients with BC, they were notably elevated in highly proliferative breast cancer subtypes, including TNBC and histological grade 3 types.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eTumor-infiltrating lymphocytes-related gene set\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTable S2 lists the enriched DEGs associated with TILs identified through ML using ANNs. The analysis revealed 49 overlapping DEGs between cohorts A (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;400) and B (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;319) (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). These 49 genes were significantly linked to gene ontologies related to immune system processes, including GO:005615 (extracellular space), GO:0072562 (blood microparticles), and GO:0031012 (extracellular matrix). In addition, Table S4 presents the top 10 molecular pathways identified among the significant pathways based on the FDR.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eThe 49 overlapping differentially expressed genes associated with tumor-infiltrating lymphocytes between cohorts A (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;400) and B (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;319)\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eEntrez Gene ID\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGene Symbol\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGene Description\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e9447\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eAIM2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAbsent in melanoma 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e64919\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eBCL11B\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBAF chromatin remodeling complex subunit BCL11B\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e51411\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eBIN2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eBridging integrator 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e6352\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCCL5\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eC\u0026ndash;C motif chemokine ligand 5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e729230\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCCR2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eC\u0026ndash;C motif chemokine receptor 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e914\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCD2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCD2 molecule\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e919\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCD247\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCD247 molecule\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e952\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCD38\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCD38 molecule\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10225\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCD96\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCD96 molecule\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1493\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCTLA4\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCytotoxic T-lymphocyte-associated protein 4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1520\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCTSS\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCathepsin S\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10663\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eCXCR6\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eC\u0026ndash;X\u0026ndash;C motif chemokine receptor 6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e115352\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eFCRL3\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFc receptor like 3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e50943\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eFOXP3\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eForkhead box P3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e2633\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eGBP1\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGuanylate-binding protein 1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3002\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eGZMB\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGranzyme B\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e29851\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eICOS\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInducible T-cell costimulator\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3594\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eIL12RB1\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInterleukin 12 receptor subunit beta 1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e8807\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eIL18RAP\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInterleukin 18 receptor accessory protein\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3561\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eIL2RG\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInterleukin 2 receptor subunit gamma\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e203522\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eINTS6L\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eIntegrator complex subunit 6 like\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3662\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eIRF4\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eInterferon regulatory factor 4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3702\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eITK\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eIL2 inducible T-cell kinase\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e27074\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eLAMP3\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLysosomal-associated membrane protein 3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e54900\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eLAX1\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLymphocyte transmembrane adapter 1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4049\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eLTA\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eLymphotoxin alpha\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e197259\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eMLKL\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMixed lineage kinase domain-like pseudokinase\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10943\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eMSL3\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMSL complex subunit 3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e84166\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eNLRC5\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNLR family CARD domain containing 5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e80380\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePDCD1LG2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eProgrammed cell death 1 ligand 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e11040\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePIM2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePim-2 proto-oncogene, serine/threonine kinase\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5450\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePOU2AF1\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePOU class 2 homeobox associating factor 1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5579\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePRKCB\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eProtein kinase C beta\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5698\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePSMB9\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eProteasome 20S subunit beta 9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e26191\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePTPN22\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eProtein tyrosine phosphatase nonreceptor type 22\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003ePTPN7\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eProtein tyrosine phosphatase nonreceptor type 7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e23231\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSEL1L3\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSEL1L family member 3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e10507\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSEMA4D\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSemaphorin 4D\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e5272\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSERPINB9\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSerpin family B member 9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4068\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSH2D1A\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSH2 domain containing 1A\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e55423\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSIRPG\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSignal-regulatory protein gamma\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e84174\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSLA2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSrc-like adapter 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e56833\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSLAMF8\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSLAM family member 8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e124460\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSNX20\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSorting nexin 20\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e11262\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eSP140\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSP140 nuclear body protein\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e201633\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eTIGIT\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eT-cell immunoreceptor with Ig and ITIM domains\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3604\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eTNFRSF9\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTNF receptor superfamily member 9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e6846\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eXCL2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eX-C motif chemokine ligand 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e79413\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cem\u003eZBED2\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eZinc finger BED-type containing 2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eHierarchical clustering was used to analyze the 49 genes based on their expression similarities (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). The cluster analysis classified the 719 patients into two subtypes: Subtype 1 (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;256, 35.6%) and subtype 2 (\u003cem\u003en\u003c/em\u003e\u0026thinsp;=\u0026thinsp;463, 64.4%). Subtype 2 exhibited a significantly higher TIL grade than subtype 1 (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001; Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eClinicopathological characteristics of subtypes 1 and 2\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFactors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u003cp\u003eTIL-related genomic subtype\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eSignificance\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSubtype 1\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSubtype 2\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTIL grade\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigh\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27 (16.6%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e136 (83.4%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003e163\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026lt;\u0026thinsp;0.0001\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eIntermediate\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e70 (31.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e154 (68.8%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e224\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eLow\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e159 (47.9%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e173 (52.1%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e332\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ePD-L1\u003c/b\u003e \u003cb\u003e(\u003c/b\u003e\u003cb\u003eCD274\u003c/b\u003e\u003cb\u003e) mRNA\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eHigh\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e67 (18.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e300 (81.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e367\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.0001\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eLow\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e189 (53.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e163 (46.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e352\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eER\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e209 (38.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e335 (61.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e5444\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.00024\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e34 (22.5%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e117 (77.5%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e151\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ePgR\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e185 (39.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e281 (60.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e466\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.00024\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e57 (25.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e167 (74.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e224\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHER2\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e34 (30.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e78 (69.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e112\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.27\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e174 (36.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e305 (63.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e479\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eSubtype\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eHR-positive and HER2-negative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e156 (40.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e230 (59.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e386\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.00034\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eHER2-positive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e34 (30.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e78 (69.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e112\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eTriple negative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e18 (19.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e75 (80.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e93\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eTumor size\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003epT2-4\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e184 (35.5%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e335 (64.5%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e519\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.93\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003epT1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e72 (36.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e128 (64.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e200\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eNodal status\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e116 (33.2%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e233 (66.8%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e349\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.27\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e136 (37.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e229 (62.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e365\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHistological grade\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eGrade 3\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e75 (25.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e222 (74.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e297\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.0001\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eGrade 1, 2\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e171 (43.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e220 (56.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e391\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAbbreviations: HR, hormonal receptor; ER, estrogen receptor; PgR, progesterone receptor; HER2, human epidermal growth factor 2; TIL, tumor-infiltrating lymphocyte\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinicopathological and prognostic significance of tumor-infiltrating lymphocyte-related gene sets\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn the cohort of 719 TCGA samples, subtype 2 was significantly associated with histological grade 3 (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), ER-negative (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.00024), PgR-negative (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.00024), and high \u003cem\u003ePD-L1\u003c/em\u003e mRNA expression (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) groups (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Interestingly, 80.6% of TNBCs were classified as subtype 2. In contrast, patients with TIL-related subtype 2 showed a significantly better prognosis than those with subtype 1 tumors (hazard ratio: 1.86, 95% confidence interval [CI]: 1.18\u0026ndash;2.93, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0075; Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). In a multivariate survival analysis, the TIL-associated genomic subtype emerged as a strong, independent adverse prognostic factor (hazard ratio: 2.27, 95% CI: 1.11\u0026ndash;4.64, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.025; Table S5).\u003c/p\u003e\u003cp\u003e\u003cb\u003eSignificance of ICOS in tumor-infiltrating lymphocyte-related gene sets\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe 49 genes in the TIL-related gene set were strongly associated with ICOS. The top interactions among the enriched TIL-related gene set identified several highly influential hubs (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), including \u003cem\u003eICOS\u003c/em\u003e, \u003cem\u003eSLA2\u003c/em\u003e, \u003cem\u003eZBED2\u003c/em\u003e, \u003cem\u003eCD96\u003c/em\u003e, \u003cem\u003eCD38\u003c/em\u003e, \u003cem\u003eGZMB\u003c/em\u003e, and \u003cem\u003eXCL1\u003c/em\u003e. Although filtered into the top interactions, these hubs were connected to numerous other interactions. The unfiltered interaction matrix was further analyzed by mathematically collapsing it into total values, leading to each gene. The genes were ranked based on total influence (Table S6), with ICOS identified as the most influential gene among the enriched genes.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eClinicopathological and prognostic significance of ICOS and its ligand at the mRNA level\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn the METABRIC cohort, high \u003cem\u003eICOS\u003c/em\u003e mRNA expression was significantly associated with high \u003cem\u003eICOSL\u003c/em\u003e mRNA expression (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), ER-negativity (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), PgR-negativity (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), human epidermal growth factor receptor 2 (HER2)-positivity (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), node-positivity (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0052), and high histological grade (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001; Table S7). Patients with low \u003cem\u003eICOS\u003c/em\u003e mRNA expression in TNBC had a worse prognosis than those with high \u003cem\u003eICOS\u003c/em\u003e mRNA expression (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Survival analysis of \u003cem\u003eICOS\u003c/em\u003e expression in other BC subtypes is shown in Figure S1. In contrast, \u003cem\u003eICOSL\u003c/em\u003e mRNA expression was not a significant prognostic factor for TNBC (Fig. S2).\u003c/p\u003e\u003cp\u003e\u003cb\u003eImmunohistochemical expression of the ICOS and its ligand protein in triple-negative BC\u003c/b\u003e\u003c/p\u003e\u003cp\u003eICOS protein expression in cancer cells was uniformly weak, whereas positive ICOS expression was observed in stromal cells, including lymphocytes surrounding the cancer cells. ICOSL protein expression was positive in the cytoplasm of cancer cells and uniformly positive in stromal cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec, d).\u003c/p\u003e\u003cp\u003eIn a cohort of 41 patients with TNBC, 27 (65.9%) were positive and 14 (34.1%) were negative for ICOS. ICOS positivity was significantly associated with high TIL (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.020) and PD-L1 positivity (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.046; Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). ICOSL expression in stromal cells was positive in 95.1% of the 41 patients with TNBC. In contrast, ICOSL expression in cancer cells was positive in 51.2%, but did not correlate with ICOS positivity in stromal cells (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eRelationship between inducible co-stimulator expression and clinicopathological features.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e\u003cp\u003eFactors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e\u003cp\u003eICOS (Stromal Cells)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eSignificance\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePositive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNegative\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eICOSL (cancer cells)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePositive\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11 (52.4%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10 (47.6%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e0.10\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e16 (80.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e4 (20.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e20\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eICOSL (stromal cells)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e26 (66.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e13 (33.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e39\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003e1.00\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e1 (50.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e1 (50.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e2\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eTIL grade\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eHigh and intermediate\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e17 (85.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e3 (15.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e20\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003e0.020\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eLow\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e10 (47.6%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e11 (52.4%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e21\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003ePD-L1 (immune cells)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive (\u003c/b\u003e\u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;\u003cb\u003e1%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e16 (84.2%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e3 (15.8%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e19\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003e0.046\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative (\u0026lt;\u0026thinsp;1%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e11 (50.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e11 (50.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e22\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eTumor size\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003epT2-4\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e14 (60.9%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e9 (39.1%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e23\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003e0.52\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003epT1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e13 (72.2%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e5 (27.8%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e18\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eNodal status\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e8 (80.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e2 (20.0%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e10\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003e0.45\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e19 (61.3%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e12 (38.7%)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e31\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003eAbbreviations: ICOS, inducible co-stimulator; ICOSL, inducible co-stimulator ligand; TIL, tumor-infiltrating lymphocyte\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eIn the 41 patients with TNBC, ICOS-positive patients tended to have better overall survival compared with ICOS-negative patients; however, this difference was not significant (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee). Conversely, ICOSL positivity in cancer cells was a significantly poor prognostic factor based on a univariate analysis (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.040; Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). Survival curves stratified by ICOS levels on stromal cells and ICOSL levels in cancer cells are shown in Figure S3.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003ePD-1 is an immune checkpoint molecule containing an immunoglobulin-like domain. It is expressed on the surface of activated lymphocytes (T and B cells) and myeloid cells [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. It binds to the PD-L1 ligand and inhibits lymphocyte activation. PD-1 is expressed in various cancer tissues and plays a major role in the immune evasion mechanisms of cancer cells [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Nivolumab and pembrolizumab are fully human monoclonal antibodies that target PD-1, whereas atezolizumab is a humanized monoclonal antibody that targets PD-L1 [\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. PD-1 and PD-L1 inhibitors have demonstrated notable antitumor effects in various cancers, including lung, gastrointestinal, melanoma, and BC, and are used as effective cancer treatments in clinical practice [\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Some immune checkpoint inhibitors exhibit high efficacy and long-term survival after treatment discontinuation, whereas others show no efficacy. Therefore, identifying biomarkers that can predict efficacy is important. At present, PD-L1 IHC expression is used as a companion diagnostic biomarker to predict the efficacy of immune checkpoint inhibitors [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]; however, guidelines for the type of PD-L1 antibody and evaluation method are lacking. In addition, assessing PD-L1 expression alone may not be sufficient to evaluate the complex tumor immune mechanisms. In contrast, TILs are thought to play an important role in tumor immunity mechanisms in cancer tissues [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Several retrospective studies have demonstrated that TIL status is valuable in predicting prognosis and response to drug therapy in cancer, making the assessment of TILs increasingly important in clinical practice [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. TILs provide information about dynamic tumor immune mechanisms and have been implicated in predicting responses to immune checkpoint inhibitors [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. TILs compromise of various immune cell types and exhibit distinct characteristic patterns associated with prognosis and prediction of response to drug therapy [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Therefore, numerous molecular biology studies have attempted to elucidate the molecular biological characteristics related to TILs [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. However, due to the complexity of the TIL pathway, additional key discoveries remain elusive. In the present study, 49 genes were found to be associated with TIL expression, showing significant correlation with TIL and PD-L1 expression. The discovery of this gene set linked to TILs may result in the identification of novel predictors of immune checkpoint inhibitor efficacy through further functional analyses. Furthermore, it may provide insights into the mechanisms of TIL induction and the intricate landscape of tumor immunity.\u003c/p\u003e\u003cp\u003eDue to the intricate molecular pathways involved in immune responses within cancer tissues, elucidating these complex mechanisms manually remains challenging. Graham \u003cem\u003eet al\u003c/em\u003e. used artificial neural network algorithms to identify candidate proliferation-related genes and evaluated their associations with clinicopathological features and outcomes in patients with BC [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. This approach allows for the simultaneous analysis of a vast array of multimodal data, including genetic information (such as DNA copy number abnormalities, RNA transcription, and protein expression), clinicopathological characteristics, and prognostic data for ML to identify predictive markers for treatment response and prognosis in BC [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. The present study integrated morphological characteristics of BC, particularly TILs, with extensive genetic information, leveraging a digital platform with ML capabilities. Digital technology was used to unravel the intricate molecular pathways controlling cancer immunity; however, challenges persist, including variability in optimal computational models and the identification of underlying biological mechanisms. Although digital technologies have been used in developing multigene profiling tools to guide chemotherapeutic treatment [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e], tools targeting tumor immunity have not yet proven clinically useful. The TIL-related gene expression patterns identified in this study enabled the classification of invasive BC into two distinct subgroups, revealing a significant difference in prognosis between these groups. This finding holds promise for the development of novel gene signatures targeting tumor immunity.\u003c/p\u003e\u003cp\u003eIn this study, the TIL-associated gene set was functionally linked to ICOS, a member of the CD28 family, such as PD-1 and cytotoxic T-lymphocyte\u0026ndash;associated antigen 4 (CTLA-4). However, ICOS is distinct as it is expressed on activated T cells [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. ICOS interacts with its specific ligand, ICOSL, which plays an important role in the differentiation of memory and effector T cells and immune responses [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. The involvement of the ICOS/ICOSL pathway in tumor growth depends on the cytokines secreted. Activation of ICOS/ICOSL leads to an increase in CD4\u0026thinsp;+\u0026thinsp;ICOS\u0026thinsp;+\u0026thinsp;and CD8\u0026thinsp;+\u0026thinsp;ICOS\u0026thinsp;+\u0026thinsp;T cells, which increases the effector T-cell to Treg ratio [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Conversely, ICOS/ICOSL signaling promotes Treg differentiation from CD4\u0026thinsp;+\u0026thinsp;T cells, which enhances the immunosuppressive effects [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. High ICOS expression in T cells is a significant biomarker for predicting clinical responses in patients with cancer treated with anti-CTLA-4 or anti-PD-1 antibodies. Mice treated with anti-ICOS antibodies showed a weak response to anti-CTLA-4 antibodies, suggesting a complex interplay between CTLA-4, PD-1, and ICOS costimulatory molecules in immune checkpoint inhibitor therapy [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Analysis of the METEBRIC cohort revealed significantly higher ICOS mRNA expression in BCs with high proliferative potential, such as high histological grade, hormone receptor negativity, and a basal subtype. In the present study, elevated ICOS expression was associated with a favorable prognosis in TNBC. In contrast, an increase in ICOS\u0026thinsp;+\u0026thinsp;Tregs in TILs has been linked to a poor prognosis in gastric cancer [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. The expression profile of ICOS/ICOSL in cancer tissues may dictate the mechanisms of tumor growth and antitumor immunity. Further studies are needed to elucidate the role of ICOS in antitumor immune responses.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe mechanism that induces TIL expression remains unclear, as TILs are a cell population with multifaceted functions. This study used ANNs to identify a set of genes associated with TILs through comprehensive RNA analysis of early-stage invasive BC.\u003c/p\u003e\u003cp\u003eFor effective therapeutic efficacy, immune checkpoint inhibitors require the presence of TILs surrounding tumors. This study aims to develop a gene panel that can facilitate the precise classification of patients who respond well to immune checkpoint inhibitors. Future studies are required to validate the utility of our TIL-related gene set in patient cohorts treated with immune checkpoint inhibitors.\u003c/p\u003e\u003cp\u003eFurthermore, our results indicated that ICOS was the most affected gene in the TIL-related gene set. Thus, the significance of ICOS in TNBC warrants further study. Understanding the intricate molecular pathways involved in cancer progression through advanced digital technologies, as demonstrated in this study, may lead to the identification of novel therapeutic targets for cancer treatment [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e].\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u003c/strong\u003e The authors gratefully acknowledge the work of our research technician, Kumiko Sudo.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Conceptualization, C.K.H., S.K., and G.R.B.; Methodology, S.K., T.F., and G.R.B.; Investigation, C.K.H., A.K., and S.K.; Original Draft Preparation, C.K.H. and S.K.; Supervision, S.O.,\u0026nbsp;T.Y., T.O., J.H., K.S., and T.F.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This research was funded by SPS KAKENHI (20K16375).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInstitutional Review Board Statement:\u0026nbsp;\u003c/strong\u003eThe study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Gunma University (HS2020-195).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent Statement:\u0026nbsp;\u003c/strong\u003ePatient consent was waived for studies using public databases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;of Interest:\u003c/strong\u003e SK has received honoraria from Kyowa Kirin Co., Ltd, Daiichi Sankyo Co., Ltd, Taiho Pharmaceutical Co., Ltd, Eli Lilly and Company, MSD K.K., AstraZeneca K.K., Chugai\u0026nbsp;Pharmaceutical, Ltd., Dinow,\u0026nbsp;Inc., Eisai Co., Ltd., Takeda Pharmaceutical Co., Ltd., and\u0026nbsp;Pfizer Japan Inc. No conflicts of interest have been reported for any of the other authors. GB is the chief scientific officer of\u0026nbsp;Intelligent Omics\u0026nbsp;Ltd, which uses the algorithms presented here to identify biological drug targets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u003c/strong\u003e Publicly available datasets were analyzed in this study. These data can be found here: https//www.cbioportal.org.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclaimer/Publisher’s Note:\u003c/strong\u003e The statements, opinions, and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions, or products referred to in the content.\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDenkert C, von Minckwitz G, Brase JC, Sinn BV, Gade S, Kronenwett R, Pfitzner BM, Salat C, Loi S, Schmitt WD, Schem C, Fisch K, Darb-Esfahani S, Mehta K, Sotiriou C, Wienert S, Klare P, Andr\u0026eacute; F, Klauschen F, Blohmer J-U, Krappmann K, Schmidt M, Tesch H, K\u0026uuml;mmel S, Sinn P, Jackisch C, Dietel M, Reimer T, Untch M, Loibl S (2015) Tumor-infiltrating lymphocytes and response to neoadjuvant chemotherapy with or without carboplatin in human epidermal growth factor receptor 2\u0026ndash;positive and triple-negative primary breast cancers. J Clin Oncol 33:983\u0026ndash;991. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1200/jco.2014.58.1967\u003c/span\u003e\u003cspan address=\"10.1200/jco.2014.58.1967\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKurozumi S, Fujii T, Matsumoto H, Inoue K, Kurosumi M, Horiguchi J, Kuwano H (2017) Significance of evaluating tumor-infiltrating lymphocytes (TILs) and programmed cell death-ligand 1 (PD-L1) expression in breast cancer. Med Mol Morphol 50:185\u0026ndash;194. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00795-017-0170-y\u003c/span\u003e\u003cspan address=\"10.1007/s00795-017-0170-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDenkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner BI, Weber KE, Budczies J, Huober J, Klauschen F, Furlanetto J, Schmitt WD, Blohmer J-U, Karn T, Pfitzner BM, K\u0026uuml;mmel S, Engels K, Schneeweiss A, Hartmann A, Noske A, Fasching PA, Jackisch C, van Mackelenbergh M, Sinn P, Schem C, Hanusch C, Untch M, Loibl S (2018) Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol 19:40\u0026ndash;50. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/s1470-2045(17)30904-x\u003c/span\u003e\u003cspan address=\"10.1016/s1470-2045(17)30904-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA (2018) profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol Biol 1711:243\u0026ndash;259. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-1-4939-7493-1_12\u003c/span\u003e\u003cspan address=\"10.1007/978-1-4939-7493-1_12\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDanaher P, Warren S, Dennis L, D\u0026rsquo;Amico L, White A, Disis ML, Geller MA, Odunsi K, Beechem J, Fling SP (2017) Gene expression markers of tumor infiltrating leukocytes. J Immunother Cancer 5:186. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/068940\u003c/span\u003e\u003cspan address=\"10.1101/068940\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLancashire LJ, Lemetre C, Ball GR (2009) An introduction to artificial neural networks in bioinformatics\u0026ndash;application to complex microarray and mass spectrometry datasets in cancer studies. Briefings Bioinf 10:315\u0026ndash;329. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bib/bbp012\u003c/span\u003e\u003cspan address=\"10.1093/bib/bbp012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbdel-Fatah TMA, Agarwal D, Liu D-X, Russell R, Rueda OM, Liu K, Xu B, Moseley PM, Green AR, Pockley AG, Rees RC, Caldas C, Ellis IO, Ball GR, Chan SYT (2016) SPAG5 as a prognostic biomarker and chemotherapy sensitivity predictor in breast cancer: a retrospective, integrated genomic, transcriptomic, and protein analysis. Lancet Oncol 17:1004\u0026ndash;1018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/s1470-2045(16)00174-1\u003c/span\u003e\u003cspan address=\"10.1016/s1470-2045(16)00174-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMian S, Ball G, Hornbuckle J, Holding F, Carmichael J, Ellis I, Ali S, Li G, McArdle S, Creaser C, Rees R (2003) A prototype methodology combining surface-enhanced laser desorption/ionization protein chip technology and artificial neural network algorithms to predict the chemoresponsiveness of breast cancer cell lines exposed to paclitaxel and doxorubicin under in vitro conditions. Proteomics 3:1725\u0026ndash;1737. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/pmic.200300526\u003c/span\u003e\u003cspan address=\"10.1002/pmic.200300526\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSmith BP, Brier ME (1996) Statistical approach to neural network model building for gentamicin peak predictions. J Pharm Sci 85:65\u0026ndash;69. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1021/js950271l\u003c/span\u003e\u003cspan address=\"10.1021/js950271l\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTan M, Deklerck R, Jansen B, Bister M, Cornelis J (2011) A novel computer-aided lung nodule detection system for CT images. Med Phys 38:5630\u0026ndash;5645. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1118/1.3633941\u003c/span\u003e\u003cspan address=\"10.1118/1.3633941\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, Cerami E, Sander C, Schultz N (2013) Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal 6:11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1126/scisignal.2004088\u003c/span\u003e\u003cspan address=\"10.1126/scisignal.2004088\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSalgado R, Denkert C, Demaria S, Sirtaine N, Klauschen F, Pruneri G, Wienert S, Van den Eynden G, Baehner FL, Penault-Llorca F, Perez EA, Thompson EA, Symmans WF, Richardson AL, Brock J, Criscitiello C, Bailey H, Ignatiadis M, Floris G, Sparano J, Kos Z, Nielsen T, Rimm DL, Allison KH, Reis-Filho JS, Loibl S, Sotiriou C, Viale G, Badve S, Adams S, Willard-Gallo K, Loi S (2014) The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs working group 2014. Ann Oncol 26:259\u0026ndash;271. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdu450\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdu450\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTong DL, Boocock DJ, Dhondalay R, Lemetre GK, Ball C GR (2014) Artificial Neural Network Inference (ANNI): A study on gene-gene interaction for biomarkers in childhood sarcomas. PLoS ONE 9:e102483. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0102483\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0102483\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ede Hoon MJL, Imoto S, Nolan J, Miyano S (2004) Open source clustering software. Bioinformatics 20:1453\u0026ndash;1454. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/bioinformatics/bth078\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/bth078\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCurtis C, Shah SP, Chin S-F, Turashvili G, Rueda OM, Dunning MJ, Speed D, Lynch AG, Samarajiwa S, Yuan Y, Gr\u0026auml;f S, Ha G, Haffari G, Bashashati A, Russell R, McKinney S, Langer\u0026oslash;d A, Green A, Provenzano E, Wishart G, Pinder S, Watson P, Markowetz F, Murphy L, Ellis I, Purushotham A, B\u0026oslash;rresen-Dale A-L, Brenton JD, Tavar\u0026eacute; S, Caldas C, Aparicio S (2012) The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486:346\u0026ndash;352. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/nature10983\u003c/span\u003e\u003cspan address=\"10.1038/nature10983\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi C, Lin L, Hou M, Chu P (2021) PD\u0026ndash;L1/PD\u0026ndash;1 blockade in breast cancer: the immunotherapy era (Review). Oncol Rep 45:5\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3892/or.2020.7831\u003c/span\u003e\u003cspan address=\"10.3892/or.2020.7831\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eParry RV, Chemnitz JM, Frauwirth KA, Lanfranco AR, Braunstein I, Kobayashi SV, Linsley PS, Thompson CB, Riley JL (2005) CTLA-4 and PD-1 receptors inhibit T-cell activation by distinct mechanisms. Mol Cell Biol 25:9543\u0026ndash;9553. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1128/mcb.25.21.9543-9553.2005\u003c/span\u003e\u003cspan address=\"10.1128/mcb.25.21.9543-9553.2005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSch\u0026uuml;tz F, Stefanovic S, Mayer L, von Au A, Domschke C, Sohn C (2017) PD-1/PD-L1 pathway in breast cancer. Oncol Res Treat 40:294\u0026ndash;297. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1159/000464353\u003c/span\u003e\u003cspan address=\"10.1159/000464353\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKeir ME, Butte MJ, Freeman GJ, Sharpe AH (2008) PD-1 and its ligands in tolerance and immunity. Annu Rev Immunol 26:677\u0026ndash;704. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1146/annurev.immunol.26.021607.090331\u003c/span\u003e\u003cspan address=\"10.1146/annurev.immunol.26.021607.090331\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBrahmer JR, Tykodi SS, Chow LQM, Hwu W-J, Topalian SL, Hwu P, Drake CG, Camacho LH, Kauh J, Odunsi K, Pitot HC, Hamid O, Bhatia S, Martins R, Eaton K, Chen S, Salay TM, Alaparthy S, Grosso JF, Korman AJ, Parker SM, Agrawal S, Goldberg SM, Pardoll DM, Gupta A, Wigginton JM (2012) Safety and activity of anti\u0026ndash;PD-L1 antibody in patients with advanced cancer. N Engl J Med 366:2455\u0026ndash;2465. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/nejmoa1200694\u003c/span\u003e\u003cspan address=\"10.1056/nejmoa1200694\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAdams S, Schmid P, Rugo HS, Winer EP, Loirat D, Awada A, Cescon DW, Iwata H, Campone M, Nanda R, Hui R, Curigliano G, Toppmeyer D, O\u0026rsquo;Shaughnessy J, Loi S, Paluch-Shimon S, Tan AR, Card D, Zhao J, Karantza V, Cort\u0026eacute;s J (2019) Pembrolizumab monotherapy for previously treated metastatic triple-negative breast cancer: cohort A of the phase II KEYNOTE-086 study. Ann Oncol 30:397\u0026ndash;404. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annonc/mdy517\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdy517\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEmens LA, Molinero L, Loi S, Rugo HS, Schneeweiss A, Di\u0026eacute;ras V, Iwata H, Barrios CH, Nechaeva M, Nguyen-Duc A, Chui SY, Husain A, Winer EP, Adams S, Schmid P (2021) Atezolizumab and \u003cem\u003enab\u003c/em\u003e-paclitaxel in advanced triple-negative breast cancer: biomarker evaluation of the IMpassion130 study. J Natl Cancer Instit 113:1005\u0026ndash;1016. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jnci/djab004\u003c/span\u003e\u003cspan address=\"10.1093/jnci/djab004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eIlson DH (2021) Adjuvant nivolumab in esophageal cancer \u0026mdash; a new standard of care. N Engl J Med 384:1269\u0026ndash;1271. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/nejme2101983\u003c/span\u003e\u003cspan address=\"10.1056/nejme2101983\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eForde PM, Spicer J, Lu S, Provencio M, Mitsudomi T, Awad MM, Felip E, Broderick SR, Brahmer JR, Swanson SJ, Kerr K, Wang C, Ciuleanu T-E, Saylors GB, Tanaka F, Ito H, Chen K-N, Liberman M, Vokes EE, Taube JM, Dorange C, Cai J, Fiore J, Jarkowski A, Balli D, Sausen M, Pandya D, Calvet CY, Girard N (2022) Neoadjuvant nivolumab plus chemotherapy in resectable lung cancer. N Engl J Med 386:1973\u0026ndash;1985. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/nejmoa2202170\u003c/span\u003e\u003cspan address=\"10.1056/nejmoa2202170\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBajorin DF, Witjes JA, Gschwend JE, Schenker M, Valderrama BP, Tomita Y, Bamias A, Lebret T, Shariat SF, Park SH, Ye D, Agerbaek M, Enting D, McDermott R, Gajate P, Peer A, Milowsky MI, Nosov A, Neif Antonio J, Tupikowski K, Toms L, Fischer BS, Qureshi A, Collette S, Unsal-Kacmaz K, Broughton E, Zardavas D, Koon HB, Galsky MD (2021) Adjuvant nivolumab versus placebo in muscle-invasive urothelial carcinoma. N Engl J Med 384:2102\u0026ndash;2114. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1056/nejmoa2034442\u003c/span\u003e\u003cspan address=\"10.1056/nejmoa2034442\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCortes J, Cescon DW, Rugo HS, Nowecki Z, Im S-A, Yusof MM, Gallardo C, Lipatov O, Barrios CH, Holgado E, Iwata H, Masuda N, Torregroza Otero M, Gokmen E, Loi S, Guo Z, Zhao J, Aktan G, Karantza V, Schmid P (2020) KEYNOTE-355 Investigators. Pembrolizumab plus chemotherapy versus placebo plus chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer (KEYNOTE-355): a randomized, placebo-controlled, double-blind, phase 3 clinical trial. Lancet 396:1817\u0026ndash;1828. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1200/jco.2020.38.15_suppl.1000\u003c/span\u003e\u003cspan address=\"10.1200/jco.2020.38.15_suppl.1000\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHonda C, Kurozumi S, Katayama A, Hanna\u0026ndash;Khalil B, Masuda K, Nakazawa Y, Ogino M, Obayashi S, Yajima R, Makiguchi T, Oyama T, Horiguchi J, Shirabe K, Fujii T (2021) Prognostic value of tumor\u0026ndash;infiltrating lymphocytes in estrogen receptor\u0026ndash;positive and human epidermal growth factor receptor 2\u0026ndash;negative breast cancer. Mol Clin Oncol 15:252. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3892/mco.2021.2414\u003c/span\u003e\u003cspan address=\"10.3892/mco.2021.2414\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEl Bairi K, Haynes HR, Blackley E, Fineberg S, Shear J, Turner S, De Freitas JR, Sur D, Amendola LC, Gharib M, Kallala A, Arun I, Azmoudeh-Ardalan F, Fujimoto L, Sua LF, Liu SW, Lien HC, Kirtani P, Balancin M, El Attar H, Guleria P, Yang W, Shash E, Chen IC, Bautista V, Do Prado Moura JF, Rapoport BL, Castaneda C, Spengler E, Acosta-Haab G, Frahm I, Sanchez J, Castillo M, Bouchmaa N, Md Zin RR, Shui R, Onyuma T, Yang W, Husain Z, Willard-Gallo K, Coosemans A, Perez EA, Provenzano E, Ericsson PG, Richardet E, Mehrotra R, Sarancone S, Ehinger A, Rimm DL, Bartlett JMS, Viale G, Denkert C, Hida AI, Sotiriou C, Loibl S, Hewitt SM, Badve S, Symmans WF, Kim RS, Pruneri G, Goel S, Francis PA, Inurrigarro G, Yamaguchi R, Garcia-Rivello H, Horlings H, Afqir S, Salgado R, Adams S, Kok M, Dieci MV, Michiels S, Demaria S, Loi S (2021) International immuno-oncology biomarker working group. The tale of TILs in breast cancer: a. npj Breast Cancer 7:150 report from the International Immuno-Oncology Biomarker Working Group\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHendry S, Salgado R, Gevaert T, Russell PA, John T, Thapa B, Christie M, van de Vijver K, Estrada MV, Gonzalez-Ericsson PI, Sanders M, Solomon B, Solinas C, Van den Eynden GGGM, Allory Y, Preusser M, Hainfellner J, Pruneri G, Vingiani A, Demaria S, Symmans F, Nuciforo P, Comerma L, Thompson EA, Lakhani S, Kim S-R, Schnitt S, Colpaert C, Sotiriou C, Scherer SJ, Ignatiadis M, Badve S, Pierce RH, Viale G, Sirtaine N, Penault-Llorca F, Sugie T, Fineberg S, Paik S, Srinivasan A, Richardson A, Wang Y, Chmielik E, Brock J, Johnson DB, Balko J, Wienert S, Bossuyt V, Michiels S, Ternes N, Burchardi N, Luen SJ, Savas P, Klauschen F, Watson PH, Nelson BH, Criscitiello C, O\u0026rsquo;Toole S, Larsimont D, de Wind R, Curigliano G, Andr\u0026eacute; F, Lacroix-Triki M, van de Vijver M, Rojo F, Floris G, Bedri S, Sparano J, Rimm D, Nielsen T, Kos Z, Hewitt S, Singh B, Farshid G, Loibl S, Allison KH, Tung N, Adams S, Willard-Gallo K, Horlings HM, Gandhi L, Moreira A, Hirsch F, Dieci MV, Urbanowicz M, Brcic I, Korski K, Gaire F, Koeppen H, Lo A, Giltnane J, Rebelatto MC, Steele KE, Zha J, Emancipator K, Juco JW, Denkert C, Reis-Filho J, Loi S, Fox SB (2017) Assessing tumor-infiltrating lymphocytes in solid tumors: a practical review for pathologists and proposal for a standardized method from the international immuno-oncology biomarkers working group: part 2: TILs in melanoma, gastrointestinal tract carcinomas, non\u0026ndash;small cell lung carcinoma and mesothelioma, endometrial and ovarian carcinomas, squamous cell carcinoma of the head and neck, genitourinary carcinomas, and primary brain tumors. Adv Anat Pathol 24:311\u0026ndash;335. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1097/pap.0000000000000161\u003c/span\u003e\u003cspan address=\"10.1097/pap.0000000000000161\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhn S, Chung YR, Seo AN, Kim M, Woo JW, Park SY (2020) Changes and prognostic values of tumor-infiltrating lymphocyte subsets after primary systemic therapy in breast cancer. PLoS ONE 15:e0233037. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0233037\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0233037\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRomagnoli G, Wiedermann M, H\u0026uuml;bner F, Wenners A, Mathiak M, R\u0026ouml;cken C, Maass N, Klapper W, Alkatout I (1936) Morphological evaluation of tumor-infiltrating lymphocytes (TILs) to investigate invasive breast cancer immunogenicity, reveal lymphocytic networks and help relapse prediction: a retrospective study. Int J Mol Sci 18:1936. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijms18091936\u003c/span\u003e\u003cspan address=\"10.3390/ijms18091936\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eL\u0026oacute;pez-Cort\u0026eacute;s A, Cabrera-Andrade A, V\u0026aacute;zquez-Naya JM, Pazos A, Gonz\u0026aacute;les-D\u0026iacute;az H, Paz-y-Mi\u0026ntilde;o C, Guerrero S, P\u0026eacute;rez-Castillo Y, Tejera E, Munteanu CR (2020) Prediction of breast cancer proteins involved in immunotherapy, metastasis, and RNA-binding using molecular descriptors and artificial neural networks. Sci Rep 10:8515. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-020-65584-y\u003c/span\u003e\u003cspan address=\"10.1038/s41598-020-65584-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWoodhouse R, Li M, Hughes J, Delfosse D, Skoletsky J, Ma P, Meng W, Dewal N, Milbury C, Clark T, Donahue A, Stover D, Kennedy M, Dacpano-Komansky J, Burns C, Vietz C, Alexander B, Hegde P, Dennis L (2020) Clinical and analytical validation of FoundationOne liquid CDx, a novel 324-Gene cfDNA-based comprehensive genomic profiling assay for cancers of solid tumor origin. PLoS ONE 15:e0237802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0237802\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0237802\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHigami A, Takada M, Kawaguchi-Sakita N, Kawashima M, Kawaguchi K, Yamaguchi A, Takeuchi Y, Yamada Y, Toi M (2024) Predicting invasive disease-free survival in ER-positive, HER2-negative early breast cancer using the PAM50 risk-of-recurrence score: a retrospective analysis using single-center long-term follow-up data of postmenopausal Japanese patients. Int J Clin Oncol. 2024;29(11):1715\u0026ndash;1720. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10147-024-02604-1\u003c/span\u003e\u003cspan address=\"10.1007/s10147-024-02604-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHutloff A, Dittrich AM, Beier KC, Eljaschewitsch B, Kraft R, Anagnostopoulos I, Kroczek RA (1999) ICOS is an inducible T-cell co-stimulator structurally and functionally related to CD28. Nature 397:263\u0026ndash;266. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/16717\u003c/span\u003e\u003cspan address=\"10.1038/16717\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSolinas C, Gu-Trantien C, Willard-Gallo K (2020) The rationale behind targeting the ICOS-ICOS ligand costimulatory pathway in cancer immunotherapy. ESMO Open 5:e000544. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/esmoopen-2019-000544\u003c/span\u003e\u003cspan address=\"10.1136/esmoopen-2019-000544\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMarinelli O, Nabissi M, Morelli MB, Torquati L, Amantini C, Santoni G (2018) ICOS-L as a Potential Therapeutic Target for Cancer Immunotherapy. Curr Protein Pept Sc 19:1107\u0026ndash;1113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2174/1389203719666180608093913\u003c/span\u003e\u003cspan address=\"10.2174/1389203719666180608093913\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhou W, Yu M, Pan H, Qiu W, Wang H, Qian M, Che N, Zhang K, Mao X, Li L, Wang R, Xie H, Ling L, Zhao Y, Liu X, Wang C, Ding Q, Wang S (2021) Microwave ablation induces Th1-type immune response with activation of ICOS pathway in early-stage breast cancer. J ImmunoTher Cancer 9:e002343. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1136/jitc-2021-002343\u003c/span\u003e\u003cspan address=\"10.1136/jitc-2021-002343\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSainson RCA, Thotakura AK, Kosmac M, Borhis G, Parveen N, Kimber R, Carvalho J, Henderson SJ, Pryke KL, Okell T, O'Leary S, Ball S, Van Krinks C, Gamand L, Taggart E, Pring EJ, Ali H, Craig H, Wong VWY, Liang Q, Rowlands RJ, Lecointre M, Campbell J, Kirby I, Melvin D, Germaschewski V, Oelmann E, Quaratino S, McCourt M (2020) An antibody targeting icos increases intratumoral cytotoxic to regulatory T-cell ratio and induces tumor regression. Cancer Immunol Res 8:1568\u0026ndash;1582. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1158/2326-6066.cir-20-0034\u003c/span\u003e\u003cspan address=\"10.1158/2326-6066.cir-20-0034\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBurlion A, Ramos RN, KC P, Sendeyo K, Corneau A, M\u0026eacute;n\u0026eacute;trier-Caux C, Piaggio E, Olive D, Caux C, Marodon G (2019) A novel combination of chemotherapy and immunotherapy controls tumor growth in mice with a human immune system. OncoImmunology 8:e1596005. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/2162402x.2019.1596005\u003c/span\u003e\u003cspan address=\"10.1080/2162402x.2019.1596005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang X, Liu X, Lin X, Yu H, Sun J, Liu X, Chen C, Jin H, Zhang G, Shi X, Zhang Q, Yu J (2014) Role of plasmacytoid dendritic cells and inducible costimulator-positive regulatory T cells in the immunosuppression microenvironment of gastric cancer. Cancer Sci 105:150\u0026ndash;158. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/cas.12327\u003c/span\u003e\u003cspan address=\"10.1111/cas.12327\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKurozumi S, Ball GR (2024) Research on biomarkers using innovative artificial intelligence systems in breast cancer. Int J Clin Oncol 29:1669\u0026ndash;1675. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10147-024-02602-3\u003c/span\u003e\u003cspan address=\"10.1007/s10147-024-02602-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Supplementary Materials","content":"\u003cp\u003eSupplementary figures S1-S3 and supplementary tables S1-S7 are not available with this version.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"tumor-infiltrating lymphocyte, machine learning, breast cancer, prognostic marker","lastPublishedDoi":"10.21203/rs.3.rs-7291997/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7291997/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe immune system plays a crucial role at all stages of tumor development, including initiation, progression, and dissemination; however, the precise molecular mechanisms underlying tumor immunity remain unclear. In this study, we aimed to identify key targets associated with tumor-infiltrating lymphocytes (TILs) in early-stage breast cancer (BC) using a novel machine learning (ML) approach. We analyzed a cohort of 719 patients with early-stage BC from The Cancer Genome Atlas datasets, all of whom had available digital hematoxylin and eosin-stained whole slide images and tumor transcriptomic data. Stromal TIL grades (low, intermediate, and high) were evaluated based on the International Working Group criteria. Using artificial neural network ML methods, we identified 49 genes that exhibited differential expression across the stromal TIL grades. Cluster analysis of these genes resulted in the classification of patients into two distinct molecular subtypes (1 and 2), which were significantly associated with tumor aggressiveness and prognosis. Our findings highlight the potential of TIL-related gene sets in deciphering the intricate molecular networks that control tumor immunity in early-stage BC.\u003c/p\u003e","manuscriptTitle":"A key molecular driver of tumor-infiltrating lymphocytes in invasive breast cancer identified by machine learning-based meta-mining","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-12 13:02:28","doi":"10.21203/rs.3.rs-7291997/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"392fbc36-b06e-47a3-9e28-2784ad89d79b","owner":[],"postedDate":"August 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-08-21T14:12:04+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-12 13:02:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7291997","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7291997","identity":"rs-7291997","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}