{"paper_id":"05a6f1ff-955d-4fe1-9563-353edf2c5d61","body_text":"Single-Cell Immune Profiling and Machine Learning Reveal a Predictive Immune Signature for Immunotherapy Response in NSCLC | 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 Single-Cell Immune Profiling and Machine Learning Reveal a Predictive Immune Signature for Immunotherapy Response in NSCLC Laura Boyero, María G Velasco-Domínguez, Alejandro Castillo-Peña, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8656976/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 Background Metastatic non-small cell lung cancer (NSCLC) exhibits high heterogeneity in response to immune checkpoint inhibitor (ICI) therapy. The identification of predictive biomarkers that are easily applicable and minimally invasive is therefore of great interest. Methods Baseline peripheral blood samples from real-world stage IV NSCLC patients treated with first-line pembrolizumab were analyzed using high-dimensional single-cell mass cytometry, clustering, and machine learning according to their response to ICI therapy. Functional immune characterization was additionally performed for validation. Results We observed significantly higher frequencies of circulating CD4⁺ and CD8⁺ EMRA T cells, defined as terminally differentiated effector memory cells re-expressing CD45RA (CD45RA⁺CCR7⁻), in non-responder patients to programmed cell death protein 1 (PD-1) inhibitors. Upon in vitro stimulation, EMRA T cell subsets from non-responders showed reduced expression of activation markers and effector molecules, including CD25, CD69, IFN-γ, and GZMB. Based on these findings, we developed the CD4⁺ EMRA ImmunoPredict Score (CEIPS), a simplified predictive model that integrates activation and regulatory markers of CD4⁺ EMRA T cells. CEIPS stratified patients into Low- and High-Risk groups, with the latter showing significantly poorer progression-free and overall survival (p = 0.005 and p < 0.001, respectively). Conclusions Peripheral CD4⁺ and CD8⁺ EMRA T cells are associated with anti-PD-1 immunotherapy response in metastatic NSCLC patients and represent clinically feasible blood-based biomarkers to improve patient stratification. Building on these findings, we developed the CEIPS score, which integrates these biomarkers and demonstrates predictive value for immunotherapy outcomes in NSCLC. Non-small cell lung cancer (NSCLC) immunotherapy immune checkpoint inhibitors pembrolizumab single-cell mass cytometry EMRA T cells peripheral blood biomarkers machine learning. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 BACKGROUND Immune checkpoint inhibitors (ICIs) have revolutionized the treatment landscape for patients with metastatic non-small cell lung cancer (NSCLC) lacking oncogenic drivers since their approval in 2015, significantly improving both survival and quality of life. However, although ICIs are effective in a subset of patients, only approximately 20% of those with NSCLC achieve durable responses and long-term clinical benefit 1 . This limited efficacy contributes to NSCLC remaining the leading cause of cancer-related death worldwide 2 , despite a gradual decline in mortality rates over the years, largely due to low screening uptake and the fact that over half of patients are diagnosed at advanced stages 3 . This underscores the urgent need for reliable predictive biomarkers to guide treatment decisions. To date, extensive efforts have been made to identify such biomarkers, including programmed cell death protein 1 ligand (PD-L1) expression 4 , tumor mutational burden (TMB) 5 , and gut microbiota composition 6 , among others. However, none have demonstrated sufficient robustness or clinical utility in the context of NSCLC. ICIs function by blocking the interaction between PD-1 (sustainably expressed on exhausted T cells) and its ligands PD-L1 and PD-L2 (commonly upregulated in the tumor microenvironment), thereby enhancing or restoring the immune response. This occurs through the promotion of cytotoxic effector activity 7–9 and/or by disrupting immune tolerance mechanisms against lung tumor cells 10 . Since ICIs primarily exert their effects by modulating T cell function, the baseline immunophenotypic landscape of peripheral blood mononuclear cells (PBMCs) may play a pivotal role in determining therapeutic response 11 , offering valuable insights into the systemic immune status of patients. In this context, advanced single-cell technologies such as mass cytometry enable high-dimensional profiling of peripheral blood, allowing in-depth functional characterization of immune populations, including states of activation, senescence, or exhaustion, that may reflect treatment responsiveness 12–14 . This approach may help identify peripheral biomarkers associated with clinical outcomes, thereby supporting personalized treatment strategies and reducing unnecessary exposure to ineffective therapies, with both clinical and economic implications 15 . In this study, we conducted comprehensive PBMC profiling and investigated biomarkers associated with clinical outcomes by integrating mass cytometry with machine learning in a real-world, unselected cohort of metastatic NSCLC patients treated with first-line pembrolizumab. Additionally, we performed in vitro T cell stimulation assays to assess functional immune characteristics under controlled conditions. Based on these findings, we developed a scoring system based on a single functional T cell subpopulation, designed for direct application in clinical settings. METHODS Patients and study design Patients were included through a search on the Virgen del Rocío Hospital (Seville, Spain) database for those diagnosed with NSCLC and who had received treatment with immunotherapy and with available follow-up data between the October 2017 and April 2021. Patient selection criteria included stage IV, having been treated with Pembrolizumab in first line, availability of a baseline sample and passing quality controls regarding viability of thawed PBMCs. A flowchart for patient selection and exclusion criteria is represented in Figure 1A. Out of a total of 145 available patients, we finally analyzed samples from those who met the eligibility criteria (n = 21). Patients received intravenous single-agent pembrolizumab at a dose of 200 mg every 3 weeks (21 days). Elegilibility for pembrolizumab treatment required PD-L1 expression levels in tumour cells to be greater than or equal to 50% in all patients. Treatment response at 6 months was evaluated in all patients with analyzed samples, according to RECIST 1.1 criteria, based on computed tomography and assessed by radiologists and medical oncologists (Fig. 1B). The objective tumor response was defined as the best clinical response observed during the course of treatment. Patients were stratified based on response into a disease control rate (DCR) group—comprising stable disease (SD), partial response (PR), and complete response (CR)—and a progressive disease (PD) group. In this study, patients classified as having disease control rate (DCR) were considered responders (R), whereas those with progressive disease (PD) were considered non-responders (NR). All participants provided written informed consent prior to enrollment, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Coordinating Committee on Biomedical Research Ethics of Andalucía (0944-N-20). Sample processing Peripheral blood samples were collected at baseline prior to pembrolizumab administration using two 8mL Cell Preparation Tubes (CPT) with sodium heparin (Vacutainer® CPT™ Mononuclear Cell Preparation Tubes, BD Biosciences). Samples were maintained at room temperature and processed within 3 hours of collection to ensure cell viability. Peripheral blood mononuclear cells (PBMC) were isolated by centrifugation following the manufacturer's instructions, washed with phosphate-buffered saline (PBS) and cryopreserved at -80°C in fetal calf serum (FCS) with 10% DMSO until processing. Quality control measures were implemented to ensure sample integrity, including viability assessment prior to cryopreservation and after thawing, before subsequent analyses. Mass cytometry After thawing, 3 million PBMCs were incubated with 20 U/μL of DNase in RPMI-1640 media for 1 hour at 37ºC and 5% CO 2 , followed by a 2-hour resting period at 37ºC in the same supplemented culture medium to allow cell recovery. Samples were included only if cell viability exceeded 85%. PBMCs were then washed and incubated with 0.5mg/mL of FcR-blocking (Human TruStain FcX, Fluidigm) for 10 minutes at room temperature (RT) to prevent nonspecific binding. Cells were subsequently stained for 30 minutes at RT using antibodies from the Maxpar Direct Immune Profiling Assay kit (Fluidigm) (Supp. Table 1), along with an additional pre-conjugated anti-CD279 (PD-1) antibody. Following surface staining, cells were permeabilized using Maxpar Perm-S Buffer (Fluidigm) and stained for 30 minutes at RT with an in-house-labeled intracellular anti-CTLA-4 antibody. All antibodies used for mass cytometry are listed in Supplementary Table 1. The in-house antibody labeling was performed using the Maxpar X8 Antibody Labeling Kit (Fluidigm), according to the manufacturer’s instructions. After washing, stained PBMCs were fixed incubating with 1.6% formaldehyde for 10 min. Cells were then incubated overnight with 125 nM of Cell-ID Intercalator-Ir solution (Fluidigm) for cell viability at 4ºC. Finally, cells were washed and stored at -80ºC until acquisition on a CyTOF2-Helios® instrument (Fluidigm) with a minimum of 200,000 events per sample. This antibody panel enabled the identification of innate immune populations, including monocytes, natural killer (NK) cells, plasmacytoid and myeloid dendritic cells (DC), and granulocytes, as well as adaptive immune populations, such as naïve, central memory, effector memory, and terminally differentiated effector memory (EMRA) CD4⁺ and CD8⁺ T cells; CD4⁺CD8⁺ and CD4⁻CD8⁻ T cells; regulatory T cells (Tregs); γδ T cells; B cells; plasmablasts; and NKT cells. In addition, functional markers including CD127, CD25, CD38, CTLA-4, HLA-DR, PD-1, CD28, and CD57 were evaluated across all of the adaptative immune subsets. Mass Cytometry Data Analysis The resulting data were processed using a pipeline that included bead-based normalization, bead exclusion, cell doublet exclusion, and live/dead cell exclusion. To identify stratifying cell clusters based on selected markers from the mass cytometry data, Uniform Manifold Approximation and Projection (UMAP) was performed using FlowJo v10.10.0 using the DownSample v3.3.1 and UMAP v4.1.1 plug-ins with default settings. First, samples were downsampled to equal numbers of events and then concatenated for analysis. The PhenoGraph v4.0.5 plug-in was used to estimate the optimal number of clusters based on the expression of multiple parameters. Subsequently, the FlowSOM v4.1.0 plug-in was applied for clustering analysis. The UMAP algorithm was run using the marker expression specified in each figure and visualized as a heatmap and density plot. Single-cell data has been clustered using the FlowSOM R package 16 and labeled using the Ek'Balam algorithm 17 . Cell subset definitions follow Maecker et al. 18 and Finak et al. 19 . Cluster labeling, method implementation, and visualization were done through the Astrolabe Cytometry Platform (Astrolabe Diagnostics, Inc.) (Supp. Table 2). Differential abundance analysis was done using the edgeR R package 20,21 following the method outlined in Lun et al. 22 . Differential expression analysis was done using the limma R package 23 following the method outlined in Weber et al. 24 Thymic output quantification Thymic output was estimated based on the δ/β T-cell receptor excision circles (TRECs) ratio, measured by droplet digital PCR (ddPCR), modified from Ferrando-Martínez et al. 25 to update the measure by droplet digital PCR (ddPCR). Briefly, genomic DNA was extracted from an average of 3×10⁶ remaining PBMCs using the QIAamp DNA Mini Kit (Qiagen), following the manufacturer’s instructions. Each ddPCR reaction, containing 150 ng of PBMC-derived DNA, was performed using highly sensitive specific probes as previously described 26 in a final volume of 20 μL using ddPCR Supermix for Probes (no dUTP, Bio-Rad) on a QX200 Droplet Digital PCR system (Bio-Rad). Data were analyzed using QuantaSoft software version 1.7.1 (Bio-Rad). Absolute telomere length Copy number quantification of the telomere repeat sequence and the single-copy gene 36B4 was performed by qPCR, following the protocol described by O’Callaghan & Fenech (2011) 27 . Each reaction contained 20 ng of genomic DNA from remaining PBMCs. Primer sequences were as follows (5′→3′): Telomere Forward (CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGGTTTGGGTT) and Reverse (GGCTTGCCTTACCCTTACCCTTACCCTTACCCTTACCCT); 36B4 Forward (CAGCAAGTGGGAAGGTGTAATCC) and Reverse (CCCATTCTATCATCAACGGGTACAA). Standard curves were generated using synthetic oligonucleotides obtained from Invitrogen for telomere sequence (TTAGGG)14 and 36B4 gene (CAGCAAGTGGGAAGGTGTAATCCGTCTCCACAGACAAGGCCAGGACTCGTTTGTACCCGTTGATGATAGAATGGG). To maintain a constant load of 20 ng per reaction, pUC19 plasmid DNA (Sigma‑Aldrich) was added to each standard tube. As a positive control, genomic DNA from the 1301 cell line (European Collection of Authenticated Cell Cultures, ECACC) was used. Reactions were run on a CFX Duet real-time PCR system (Bio-Rad) and absolute telomere length were expressed as kilobases of telomeric repeats per single-copy gene (kb/SCG), as previously described 27 . Flow cytometry PBMCs were isolated as previously described, thawed, and incubated with Benzonase (Sigma-Aldrich) for 15 minutes at 37°C. PBMCs (5×10⁶ cells/mL) were then cultured in RPMI-1640 medium (Invitrogen Life Technologies) supplemented with human serum (Sigma-Aldrich) and antibiotics (Gibco), and stimulated with 5 μg/mL phytohemagglutinin (PHA, Thermo Fisher) for 24 hours at 37°C, in a 5% CO₂ atmosphere. To block cytokine secretion, GolgiStop (BD Biosciences) and Brefeldin A (BioLegend) were added to the cultures for 4 hours, following the manufacturers’ instructions. Cells were washed and resuspended in DPBS (Gibco) at a concentration of 2×10⁶ cells/mL, and stained with Fixable Dead Cell Yellow (Invitrogen) at a 1:10,000 dilution for 30 minutes in the dark. Subsequently, cells were washed, fixed, and permeabilized using Cytofix/Cytoperm reagent (BD Biosciences) for intracellular staining, incubated for 20 minutes at 4°C in the dark. Finally, cells were stained overnight at 4°C in a wash buffer solution containing saponin, Brilliant Stain Buffer (BD Biosciences), and Fc Block (BD Biosciences), using pre-titrated fluorochrome-conjugated monoclonal antibodies: anti-CD3-Alexa700 (1:5000, Invitrogen), anti-CD4-APC-Cy7 (1:5000, BD Biosciences), anti-CD8-BV711 (1:100, Invitrogen), anti-CCR7-RB613 (1:5000, BD Biosciences), anti-CD45RA-PE-Cy7 (1:4000, Invitrogen), anti-CD25-BV605 (1:20, BD Biosciences), anti-CD69-APC (1:1000, BD Biosciences), anti-IFN-γ-BV421 (1:1000, BioLegend) and anti-GZMB-FITC (1:25, BD Biosciences). Data acquisition was performed using a Cytek Aurora spectral cytometer (Cytek Biosciences), and data were analyzed with FlowJo software (v10.8.1). Statistical analyses Data are presented as medians with interquartile ranges (IQRs). Statistical comparisons between groups were conducted using the non-parametric Mann-Whitney U test for continuous variables. All statistical analyses and data visualizations were performed using IBM SPSS Statistics v26.0, GraphPad Prism v6.01 (GraphPad Software, Inc.), or the SRplot platform (http://www.bioinformatics.com.cn) 28 , employing non-parametric statistical tests in all cases. Outliers were identified and excluded using the ROUT method (Q = 1%). Survival analyses were performed by Kaplan-Meier method and comparisons between groups were performed using the log-rank test. Progression-free survival (PFS) was defined as the time from the initiation of pembrolizumab treatment to either clinical or radiographic disease progression, death from any cause, or censoring at the data cut-off date. Overall survival (OS) was defined as the time from the start of pembrolizumab treatment to death from any cause or censoring at the data cut-off. Survival curves were plotted using the survminer package in R, displaying 95% confidence intervals (CIs). A p-value < 0.05 was considered statistically significant. Machine Learning-Based Predictive Model To develop a predictive model of response to first-line pembrolizumab in stage IV NSCLC based on differential marker expression in T cell subpopulations, we employed Random Forest, a supervised machine learning classification algorithm, to identify the most informative variables. This analysis was performed using the caret package in R v4.3.2. Since no independent dataset was available, model performance was validated using bootstrapping, a widely accepted resampling technique that helps mitigate overfitting and assess model stability in small datasets. Variable importance was calculated based on the Gini impurity index. A simplified model was then constructed using the top-ranked variables. Model performance metrics included the area under the curve (AUC), sensitivity, specificity, precision, and Cohen’s kappa coefficient. The ROC curve was plotted using the ROCR package in R. RESULTS Patient Characteristics and Clinical Outcomes Baseline clinicopathological and demographic characteristics of the patients are summarized in Table 1. A total of 21 subjects (aged 57 to 80 years) with stage IV NSCLC who received first-line pembrolizumab treatment were analyzed and categorized based on their treatment response at 6 months, according to RECIST 1.1 criteria. This resulted in n=10 (47.6%) responders (R) and n=11 (52.4%) non-responders (NR). The median age of the patients was 67 years (range 57 to 80). The majority of patients were male (85.7%). Regarding performance status, 30.4% had an ECOG-PS score of 0, 47.8% had a score of 1, and 14.3% had a score of 2. The cohort was predominantly composed of patients with adenocarcinoma histology (81%). Regarding smoking history, over 70% of patients were either current smokers or had quit smoking within the last 10 years. The distribution patterns for age, sex, ECOG performance status, histological subtype, and smoking habits were comparable between the two treatment response groups, demonstrating no significant correlation between these characteristics and the response groups (p > 0.05; Table 1). The median follow-up duration for the entire cohort was 25.23 months (95% CI 21.16-46.43). Patients in the R group had significantly better PFS and OS compared to those in the NR group (p < 0.0001). Specifically, the median PFS and OS in the R group were 55.53 (95% CI 35.92-61.62) and 62.8 (95% CI 54.62-70.17) months, respectively, whereas in the NR group, the median PFS and OS were 3.5 (95% CI 2.07-4.93) and 13.9 months (95% CI 5.9-22.67) (Fig. 1C, D; Supp. Table 3). Table 1 . Baseline patient characteristics. ECOG, Eastern Cooperative Oncology Group. High-Dimensional Immune Profiling of Peripheral Blood Baseline peripheral blood samples from patients with metastatic NSCLC treated with pembrolizumab were analyzed using mass cytometry to identify immune signatures associated with clinical response. To visualize and explore the high-dimensional single-cell data, we applied UMAP, a machine learning-based dimensionality reduction method, which revealed distinct clustering of immune cell subsets that differed between R and NR (Fig. 1E). We identified the major immune cell types—including T lymphocytes (CD3 + ), B lymphocytes (CD19 + ), natural killer (NK) cells (CD56 + ), granulocytes (CD66b + ), dendritic cells (DCs, CD11c + ), and monocytes (CD14 + )—through manual gating based on the expression of key phenotypic markers (Fig. 1F). We then compared the relative abundance of each major immune cell type between R and NR, and observed a trend toward a lower proportion of T and NK cells and a higher prevalence of monocytes, granulocytes, and dendritic cells (DCs) in R group compared to NR; however, these differences did not reach statistical significance (p > 0.05) (Fig. 1G). T Cell Subset Characterization and Clustering Analysis Given the observed difference in the relative abundance of CD3⁺ T cells between NR and R (fold change [FC] ~1.5), and considering the central role of T cells in the mechanism of action of immunotherapy, we sought to perform an in-depth characterization of T cell subpopulations to explore their potential as predictive biomarkers of treatment efficacy. To this end, we gated CD3⁺ T cells and generated a new UMAP to visualize the clustering patterns of T cell subsets (Fig. 2A; Supp. Fig. 1). We then manually gated various T cell subsets, including CD4⁺ T cells (Fig. 2B, C), which encompassed naïve (CD45RA⁺ CCR7⁺), central memory (CD45RA⁻ CCR7⁺), effector memory (CD45RA⁻ CCR7⁻), and EMRA cells—terminally differentiated effector memory cells re-expressing CD45RA (CD45RA⁺ CCR7⁻)—as well as regulatory T cells (Tregs; CD25⁺). Similarly, CD8⁺ T cells (Fig. 2D, E) were classified according to the same differentiation stages. In addition, we identified non-classical CD4/CD8 T cell subsets (Supp. Fig. 2A, B), comprising γδ T cells (TCRγδ⁺), double-positive T cells (CD4⁺ CD8⁺), and double-negative T cells (CD4⁻ CD8⁻). Comparison of T cell subset frequencies revealed a higher proportion of CD4⁺ EMRA cells in the NR group compared to R (FC = 2.82), along with an increased frequency of CD8⁺ EMRA cells (FC = 1.24), indicating that both CD4⁺ and CD8⁺ EMRA subsets were enriched in NR patients (Fig. 2B, D). Conversely, the R group showed higher frequencies of CD4⁺ central memory T cells, CD8⁺ central memory T cells, and CD8⁺ effector memory T cells compared to NR (FC = 1.29, 1.50, and 1.46, respectively). No enrichment was observed between NR and R in the frequencies of naïve T cells, regulatory T cells (Tregs), or non-classical CD4/CD8 subsets. Given the compositional differences observed across T cell subsets, we next explored whether they might reflect underlying alterations in T cell ontogeny and maturation, processes that primarily occur in the thymus, and could indicate a weakened immune system in NR patients. To address this question, we quantified thymic output within the T cell compartment by comparing the δ/β T-cell receptor excision circles (TRECs) ratio and absolute telomere length between R and NR patients, using a limited number of samples for which remaining PBMCs were available. The NR group showed a lower thymic output than the R group, with median δ/β TREC ratios of 0.4706 and 1.567, respectively (Fig. 3A). Although this difference did not reach statistical significance (p = 0.151), likely due to the limited sample size, exclusion of a single outlier identified by the ROUT method (Q = 1%) revealed a significant difference (p = 0.016). These findings suggest that R patients may retain greater thymic function and benefit from a broader and more diverse TCR repertoire, potentially contributing to their improved immune response. In contrast, the baseline for the telomere absolute length in circulating mononuclear cells showed no significant differences between response groups (Supp. Fig. 2C). To complement and validate these findings through an unbiased approach, we performed unsupervised clustering on the CD3⁺ T cell population using the PhenoGraph and FlowSOM algorithms. This analysis, followed by manual annotation, allowed the identification of 26 distinct T cell subpopulations based on similarities in the expression profiles of multiple immune markers. Among these clusters, we identified all major T cell subsets, including CD4⁺ and CD8⁺ naïve, central memory, effector memory, and EMRA cells, as well as Tregs, double-positive (CD4⁺CD8⁺), double-negative (CD4⁻CD8⁻), and γδ T cells (Fig. 3B; Supp. Table 4). Additionally, several subpopulations exhibited expression of activation, exhaustion and senescence markers (e.g., CD38, HLA-DR and CD57). A qualitative comparison of the identified T cell subpopulations between R and NR groups revealed higher frequencies of CD4⁺ EMRA, CD8⁺ EMRA, and CD4⁺CD8⁺ T cells in the NR group (FC = 3.66, 2.07, and 2.10, respectively). In parallel with manual gating and qualitative analysis of CD3⁺ T cell subpopulations, an automated de-multiplexed algorithm was applied using the Astrolabe platform (Astrolabe Diagnostics Inc., Arlington, VA, USA). This algorithm implements a standardized analysis pipeline to quantitatively assess pre-defined immune cell subsets and identify statistically significant differences between groups. For this analysis, only CD3⁺ T cell populations were included in the automated workflow to compare R and NR (Fig. 3C, Supp. Fig. 2D). Significant differences in the EMRA T cell compartment between R and NR were identified based on the output of the Astrolabe analysis. Specifically, CD4⁺ EMRA and CD8⁺ EMRA subsets were found to be significantly increased in NR, with Log 2 FC of -1.94 and -1.38, and corresponding -Log₁₀ (FDR) values of 1.75 and 1.01, respectively (Fig. 3D). These results reinforce the enrichment of terminally differentiated T cell subsets in NR patients, particularly within the EMRA compartment, as revealed consistently by both manual and automated analyses. Functional Profiling of EMRA T Cells We also performed an in-depth analysis of baseline differential expression patterns of activation/regulatory and exhaustion/senescence markers across CD3⁺ T cell subpopulations, comparing R and NR using the standardized pipeline of the Astrolabe platform. Specifically, we evaluated the median FC in the expression of CD127, CD25, CD38, CTLA-4 and HLA-DR as activation/regulatory markers, and PD-1, CD28, and CD57 as exhaustion/senescence-associated markers. While no significant differences were observed for exhaustion/senescence markers, the R group exhibited significantly higher expression of CTLA-4 and HLA-DR in several CD3⁺ T cell subsets, with CD4⁺ EMRA cells showing the most pronounced differences (p = 0.007 and 0.002, respectively) (Fig. 3E; Supp. Table 5). Given these findings, we sought to assess the actual functional activation capacity of the EMRA T cell compartment in R and NR patients. To this end, we conducted an in vitro stimulation assay in a limited subset of patients with available PBMCs (n = 8; 4 R and 4 NR), stimulated ex vivo with phytohemagglutinin (PHA), a potent T cell activator. Spectral flow cytometry was used to identify CD4⁺ and CD8⁺ EMRA T cells (CD45RA⁺CCR7⁻) (Fig. 4A). This analysis confirmed a trend toward a higher frequency of EMRA T cells in NR patients, with median values of 4.99% (NR) vs. 3.17% (R) for CD4⁺ EMRA T cells and 32.65% (NR) vs. 25.90% (R) for CD8⁺ EMRA T cells (Fig. 4B), although these differences did not reach statistical significance. Further analysis of effector molecules and activation markers, including CD25, CD69, IFN-γ, and granzyme B (GZMB), revealed higher expression levels in both CD4⁺ (Fig. 4C, Supp. Fig. 3A) and CD8⁺ EMRA T cells (Fig. 4D, Supp. Fig. 3B) from R patients compared to NR. These results suggest that R patients harbor a smaller EMRA T cell compartment, yet with greater functional capacity, than NR patients. Development of a Predictive Model for Pembrolizumab Response We sought to develop a robust and clinically relevant model to predict pembrolizumab efficacy in NSCLC. To minimize complexity and enhance translational applicability, we preselected variables that showed significant differences between R and NR groups; specifically, the relative abundance of CD4⁺ EMRA and CD8⁺ EMRA T cell subsets, along with the expression levels of key activation/regulatory markers (CD25, CTLA-4, and HLA-DR) within these populations. Given the relevance of the functional balance between activation and inhibition, we also included biologically meaningful expression ratios, namely CD25/CTLA-4 and HLA-DR/CTLA-4, calculated within both CD4⁺ and CD8⁺ EMRA subsets. These ratios were hypothesized to reflect shifts in T cell functional states with potential predictive value. To identify the most informative predictors and further reduce the variable set for the final model, we applied the non-parametric machine learning algorithm Random Forest to rank the contribution of each feature to classification performance (Fig. 5A). In addition to immunological features, we also included relevant clinical variables, such as age, sex, ECOG performance status, histology and smoking status, in the initial model. However, the relative contribution of these clinical variables was found to be ≤5%, and they were therefore excluded from the final model (Supp. Fig. 4). The Random Forest analysis identified the CD25/CTLA-4 ratio in CD4⁺ EMRA T cells, along with the expression levels of CTLA-4 and HLA-DR within the same subset, as the top contributors to the predictive model. We independently evaluated the predictive and prognostic value of the three top-ranked parameters identified. All three variables (CD25/CTLA-4 ratio, CTLA-4 expression, and HLA-DR expression in CD4⁺ EMRA T cells) showed statistically significant differences between R and NR groups, with higher values observed in responders (Fig. 5B, C, D). Among them, the strongest association with response was observed for the CD25/CTLA-4 ratio (p < 0.0001), followed by HLA-DR expression (p = 0.006) and CTLA-4 expression (p = 0.010). Kaplan-Meier survival analysis further demonstrated that elevated levels of all three parameters were significantly associated with improved PFS (p = 0.001 for CD25/CTLA-4 ratio, p = 0.002 for HLA-DR, and p = 0.023 for CTLA-4). In terms of OS, high levels of the CD25/CTLA-4 ratio (p = 0.007) and HLA-DR expression (p = 0.022) were associated with better outcomes, whereas CTLA-4 expression showed a similar trend but did not reach statistical significance (p = 0.091) (Fig. 5E). Although each of these parameters independently demonstrated predictive value for treatment response and survival, we investigated whether their combination could enhance overall predictive performance. Given the limited sample size, we employed bootstrap cross-validation within our cohort. This approach involved 100 random resampling iterations from the original dataset to minimize overfitting bias. The combined model, integrating the three CD4⁺ EMRA-associated parameters, effectively predicted response, achieving an area under the curve (AUC) of 0.908, with a sensitivity of 0.832 and a specificity of 0.869 (Fig. 6A). These findings support the robustness of the model and indicate that the observed results are not driven by a small number of outliers. CD4⁺ EMRA ImmunoPredict Score (CEIPS) Stratifies Clinical Outcomes To facilitate future clinical translation of the predictive model, we developed a simplified scoring system, termed the CD4⁺ EMRA ImmunoPredict Score (CEIPS) . One point was assigned for each of the three model variables (CD25/CTLA-4 ratio, CTLA-4 expression, and HLA-DR expression in CD4⁺ EMRA T cells) when their values fell below the corresponding median, resulting in a cumulative score ranging from 0 to 3. Higher scores in this CEIPS were associated with poorer prognosis. All patients with a score of 0 (S0) were classified within the R group, while scores of 3 (S3) were exclusively observed in the NR group. In addition, scores of 1 (S1) were more frequent in R, whereas scores of 2 (S2) predominated in NR (Fig. 6B). Based on this distribution, scores of 0 or 1 were categorized as Low Risk, and scores of 2 or 3 as High Risk. Kaplan-Meier analysis revealed significantly poorer PFS and OS in the High-Risk group compared to the Low-Risk group (p = 0.005 and p < 0.001, respectively; Fig. 6C, D). These findings are further illustrated in Figures 6E and 6F, which show the individual PFS and OS outcomes for patients stratified by the CEIPS classification. DISCUSSION We analyzed the comprehensive PBMC profile of real-world metastatic NSCLC patients treated with first-line pembrolizumab, with treatment response assessed at 6 months and a median follow-up of 25.23 months. By combining single-cell mass cytometry, clustering and machine learning, and functional immune characterization, we developed a simple predictive model (CEIPS) based on a single functional T cell subpopulation. This integrative approach captures complementary immunological features, including early activation (CD25), sustained activation (HLA-DR), and immune regulation (CTLA-4), within the CD4⁺ EMRA subset, thereby providing a multidimensional perspective on T cell functionality relevant to immunotherapy response. The resulting score is clinically feasible and readily translatable, enabling real-time monitoring of ICI treatment response through noninvasive liquid-biopsy analysis. Despite the clinical success of ICIs across several tumor types 29 , including NSCLC 30,31 , only a minority of patients (~20%) 32 derive durable benefit. As clinical indications continue to expand, an increase in the proportion of non-responders is expected. Proposed biomarkers such as PD-L1 expression, microsatellite instability-high (MSI-H) status, and tumor mutational burden (TMB) have shown limited predictive efficacy 33 , underscoring the need for functional immune readouts that more accurately reflect host-tumor interactions. In this context, single-cell and machine learning approaches have enabled deeper dissection of immune heterogeneity, yet their clinical translation remains limited. Blood-based biomarkers, however, offer a readily accessible source of longitudinal information with the potential to complement or even surpass tumor-based biomarkers. In line with this landscape, our study identifies an immune signature with predictive value in NSCLC with immune checkpoint inhibitors, providing a liquid-biopsy-based framework for treatment monitoring. Our analysis of peripheral immune cell subpopulations revealed a lower proportion of CD3⁺ T lymphocytes in responders compared with non-responders at baseline.Notably, similar findings have been reported in melanoma 34 . Likewise, we observed an association between the circulating EMRA T cell compartment (CCR7⁻ CD45RA⁺) and treatment responsiveness, with responders displaying lower baseline frequencies of both CD4⁺ EMRA and CD8⁺ EMRA subsets. Independent data from melanoma 35,36 and mesothelioma 37 support these observations.Although seemingly counterintuitive, one plausible explanation is that responders harbor T cells with enhanced migratory capacity toward the tumor, where they infiltrate and exert antitumor activity 34 . This redistribution of T cells from blood to tumor increases the number of tumor-infiltrating lymphocytes (TILs) in responders 38,39 , a hallmark of “hot” tumors 40 , and may indirectly contribute to the elevated frequency of circulating myeloid cells consistently reported in previous studies 34,41–43 . Similarly, the peripheral increase in CD4⁺ and CD8⁺ EMRA clusters (T cells representing the most differentiated stage 44 ) observed in non-responders may be attributable to differences in trafficking to the tumor, where they constitute only a small fraction of tumor-infiltrating T cells 45 . Indeed, EMRA T cell clusters exhibit the highest migration indices between blood and tumor tissues across most cancer types, and increased infiltration of CD8⁺ EMRA cells has been reported in some tumor samples from lung cancer and melanoma, and more consistently in renal cancer 46 , highlighting variability among cancer types. Beyond migration, EMRA clusters may also act as direct antitumor mediators. Supporting this notion, although EMRA T cells were less frequent in responders, they displayed greater in vitro functional activity, with higher expression of activation markers (CD25 and CD69) as well as effector molecules (IFN-γ and GZMB). Moreover, they showed increased expression of HLA-DR, a molecule associated with T cell activation and proliferation 47 , together with elevated CTLA-4 expression, typically linked to T cell exhaustion 48 in chronic conditions such as cancer, but also a key regulator of adaptive immunity 49 . This pattern is consistent with activation-induced inhibitory feedback and may reflect a highly activated but regulated T cell state, which may support initial response while potentially predisposing to immune exhaustion and resistance to anti-PD-1 monotherapy, thereby providing a rationale for combined PD-1/CTLA-4 blockade. Collectively, these data suggest that EMRA T cells from responders possess enhanced antitumor potential, dependent on their expressed gene signature. On this basis, we incorporated into the development of the CEIPS model not only the abundance of EMRA subsets but also their expression levels of key activation and regulatory markers. It is worth noting that the EMRA T cell pool increases with age and has been associated with immunosenescence 44 .Moreover, EMRA cells can be subdivided according to CD57 expression, with functional and proliferative (CD57⁻) subsets and senescent (CD57⁺) subsets, the latter characterized by increased telomere shortening and reduced TREC content 50–52 . In our cohort, no significant age differences were found between responders and non-responders, suggesting that the differences observed in EMRA abundance may instead be attributable to reduced thymic output in non-responders, reflected in low δ/β TREC levels. This could limit the generation of new, potentially tumor-reactive TCRs, thereby favoring compensatory clonal expansion of preexisting populations in non-responders. Nevertheless, further studies will be needed to clarify this issue. Whether these clonotypes are specific for tumor antigens remains controversial, although supportive evidence has been reported in circulating EMRA cells from breast cancer patients 53,54 . This study faced limitations typical of mass cytometry, such as a reduced cohort size due to the high cost of the technique and the challenge of high-dimensional data analysis 55 . The dimensionality reduction algorithm UMAP provides a qualitative overview but does not yield statistical analyses. Meanwhile, the Astrolabe pipeline streamlines data processing but offers limited standardization with relatively rigid analyses. Nevertheless, this automation helps to minimize operator bias and subjectivity. To mitigate these issues, supervised manual gating and statistical analyses were also implemented, along with functional assays conducted by experienced scientists to enable a deeper exploration of the data. On the other hand, this study is strengthened by the use of a real-world cohort of patients not selected by strict and complex enrollment criteria typically applied in clinical trials 56 , which often limit generalizability 57 . Furthermore, it adopts a translational approach by developing a clinically applicable score, simplified to a few parameters within a single T cell population, which can be monitored using small flow cytometry panels. Validation performed through machine learning and bootstrapping further supports the robustness of the model. CONCLUSIONS In conclusion, we identified circulating CD4⁺ and CD8⁺ EMRA T cells as predictors of response to first-line pembrolizumab therapy in metastatic NSCLC. We further developed the CD4⁺ EMRA ImmunoPredict Score (CEIPS), a predictive model based on activation and regulatory markers of CD4⁺ EMRA T cells that showed strong predictive capacity. These findings provide a noninvasive and feasible tool that could aid therapeutic decision-making in the context of personalized medicine and may also contribute to disease monitoring. Nonetheless, validation in larger prospective cohorts and across other tumor types will be required. If confirmed, this strategy could pave the way toward clinically implementable immune-functional biomarkers that complement or surpass current tumor-based assays. Declarations Availability of data and materials Data are available in a public, open access repository. Data supporting this publication are available at ImmPort (https://www.immport.org) under study accession SDY3250. Ethics approval and consent to participate All participants provided written informed consent prior to enrollment, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Coordinating Committee on Biomedical Research Ethics of Andalucía (0944-N-20). Contributions L.B. conceived the study, provided conceptualization, analyzed and discussed the results, and wrote the manuscript. M.G.V.-D., S.B.-C., and S.L.-M. performed and validated the experimental data. M.G.V.-D. and A.C.-P. conducted the statistical analyses. A.C.-P., L.B., and M.A.M.-F. developed the predictive model. J.C.B., M.A.-G., A.S.-G., and R.B. were responsible for patient recruitment, and J.C.B. evaluated clinical responses. S.M.-P., Y.M.P., and R.B. provided supervision, scientific expertise, and critical feedback. S.M.-P. and R.B. participated in funding acquisition. All authors reviewed and approved the final manuscript. Funding information The authors declare financial support was received for the research and/or publication of this article. LB and MAG were funded by Regional Ministry of Health and Consume of Andalucía (PI-0196-2025). AC-P was supported by an FPU22/04225 fellowship funded by the Spanish Ministry of Education. YMP was supported by grant CNS2023-144725 funded by AEI. SM-P was funded by the Ministry of Health and Social Welfare of Junta de Andalucia (Nicolás Monardes Program RC1-0005-2025), ISCIII (PI23/01679) and co-funded by FEDER from Regional Development European Funds (European Union). Acknowledgments We would like to acknowledge patients and their families for donating biological samples. Competing interests The authors declare that they have no competing interests. References Topalian, S. L., Taube, J. M., Anders, R. A. & Pardoll, D. M. Mechanism-driven biomarkers to guide immune checkpoint blockade in cancer therapy. Nat Rev Cancer 16 , 275 (2016). Siegel, R. L., Kratzer, T. B., Giaquinto, A. N., Sung, H. & Jemal, A. Cancer statistics, 2025. Ca 75 , 10 (2025). Olivares-Hernández, A. et al. Immune checkpoint inhibitors in non-small cell lung cancer: from current perspectives to future treatments—a systematic review. Ann Transl Med 11 , 354–354 (2023). Kluger, H. M. et al. PD-L1 Studies Across Tumor Types, its Differential Expression and Predictive Value in Patients Treated with Immune Checkpoint Inhibitors. Clin Cancer Res 23 , 4270 (2017). Prasad, V. & Addeo, A. The FDA approval of pembrolizumab for patients with TMB >10 mut/Mb: was it a wise decision? No. Annals of Oncology 31 , 1112–1114 (2020). Grenda, A. et al. Gut microbial predictors of first-line immunotherapy efficacy in advanced NSCLC patients. Scientific Reports 2025 15:1 15 , 1–12 (2025). Zhou, Q. et al. Blockade of Programmed Death-1 Pathway Rescues the Effector Function of Tumor-Infiltrating T Cells and Enhances the Antitumor Efficacy of Lentivector Immunization. The Journal of Immunology 185 , 5082–5092 (2010). Blank, C. et al. Blockade of PD-L1 (B7-H1) augments human tumor-specific T cell responses in vitro. Int J Cancer 119 , 317–327 (2006). Wong, R. M. et al. Programmed death-1 blockade enhances expansion and functional capacity of human melanoma antigen-specific CTLs. Int Immunol 19 , 1223–1234 (2007). Shepherd, F. A., Douillard, J. Y. & Blumenschein, G. R. Immunotherapy for Non-small Cell Lung Cancer: Novel Approaches to Improve Patient Outcome. Journal of Thoracic Oncology 6 , 1763–1773 (2011). Rochigneux, P. et al. Medical Treatment of Lung Cancer: Can Immune Cells Predict the Response? A Systematic Review. Front Immunol 11 , 1036 (2020). Caligola, S., De Sanctis, F., Canè, S. & Ugel, S. Breaking the Immune Complexity of the Tumor Microenvironment Using Single-Cell Technologies. Front Genet 13 , (2022). Christodoulou, M. I. & Zaravinos, A. Single-Cell Analysis in Immuno-Oncology. Int J Mol Sci 24 , (2023). Nikanjam, M., Kato, S. & Kurzrock, R. Liquid biopsy: current technology and clinical applications. J Hematol Oncol 15 , 131 (2022). Seliger, B. & Massa, C. CyTOF as a suitable tool for stratification and monitoring of cancer patients. J Transl Med 23 , 724 (2025). Van Gassen, S. et al. FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data. Cytometry Part A 87 , 636–645 (2015). Amir, E. ad D. et al. Development of a comprehensive antibody staining database using a standardized analytics pipeline. Front Immunol 10 , 458295 (2019). Maecker, H. T., McCoy, J. P. & Nussenblatt, R. Standardizing immunophenotyping for the Human Immunology Project. Nat Rev Immunol 12 , 191–200 (2012). Finak, G. et al. Standardizing Flow Cytometry Immunophenotyping Analysis from the Human ImmunoPhenotyping Consortium. Sci Rep 6 , (2016). Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 , 139–140 (2010). McCarthy, D. J., Chen, Y. & Smyth, G. K. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucleic Acids Res 40 , 4288–4297 (2012). Lun, A. T. L., Richard, A. C. & Marioni, J. C. Testing for differential abundance in mass cytometry data. Nat Methods 14 , 707–709 (2017). Ritchie, M. E. et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43 , e47 (2015). Weber, L. M., Nowicka, M., Soneson, C. & Robinson, M. D. diffcyt: Differential discovery in high-dimensional cytometry via high-resolution clustering. Commun Biol 2 , (2019). Ferrando-Martínez, S. et al. A reliable and simplified sj/β-TREC ratio quantification method for human thymic output measurement. J Immunol Methods 352 , 111–117 (2010). Bulnes-Ramos, Á. et al. Factors associated with the humoral response after three doses of COVID-19 vaccination in kidney transplant recipients. Front Immunol 14 , 1099079 (2023). O’Callaghan, N. J. & Fenech, M. A quantitative PCR method for measuring absolute telomere length. Biol Proced Online 13 , 3 (2011). Tang, D. et al. SRplot: A free online platform for data visualization and graphing. PLoS One 18 , e0294236 (2023). Zhang, Y. & Zhang, Z. The history and advances in cancer immunotherapy: understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. Cell Mol Immunol 17 , 807–821 (2020). Tan, S., Day, D., Nicholls, S. J. & Segelov, E. Immune Checkpoint Inhibitor Therapy in Oncology: Current Uses and Future Directions: JACC: CardioOncology State-of-the-Art Review. JACC CardioOncol 4 , 579–597 (2022). Ömeroğlu, G. & Mşek, Ş. Immunotherapy in the Treatment of Cancer: Today and Tomorrow. Current Molecular Biology Reports 2024 10:3 10 , 54–64 (2024). Prasad, V., Haslam, A. & Olivier, T. Updated estimates of eligibility and response: Immune checkpoint inhibitors. Journal of Clinical Oncology 42 , e14613–e14613 (2024). Holder, A. M. et al. Defining clinically useful biomarkers of immune checkpoint inhibitors in solid tumours. Nature Reviews Cancer 2024 24:7 24 , 498–512 (2024). Krieg, C. et al. High-dimensional single-cell analysis predicts response to anti-PD-1 immunotherapy. Nature Medicine 2018 24:2 24 , 144–153 (2018). Subrahmanyam, P. B. et al. Distinct predictive biomarker candidates for response to anti-CTLA-4 and anti-PD-1 immunotherapy in melanoma patients. J Immunother Cancer 6 , 18 (2018). Wistuba-Hamprecht, K. et al. Peripheral CD8 effector memory type 1 T-cells correlate with outcome in ipilimumab-treated stage IV melanoma patients. Eur J Cancer 73 , 61 (2017). Mankor, J. M. et al. Efficacy of nivolumab and ipilimumab in patients with malignant pleural mesothelioma is related to a subtype of effector memory cytotoxic T cells: Translational evidence from two clinical trials. EBioMedicine 62 , 103040 (2020). Kluger, H. M. et al. Characterization of PD-L1 expression and associated T-cell infiltrates in metastatic melanoma samples from variable anatomic sites. Clinical Cancer Research 21 , 3052–3060 (2015). Daud, A. I. et al. Tumor immune profiling predicts response to anti–PD-1 therapy in human melanoma. J Clin Invest 126 , 3447–3452 (2016). Wang, L. et al. Hot and cold tumors: Immunological features and the therapeutic strategies. MedComm (Beijing) 4 , e343 (2023). Jeon, S. H. et al. Dynamic changes in peripheral blood monocytes early after anti-PD-1 therapy predict clinical outcomes in hepatocellular carcinoma. Cancer Immunol Immunother 72 , 371 (2022). Zhou, J. G. et al. Prospective development and validation of a liquid immune profile-based signature (LIPS) to predict response of patients with recurrent/metastatic cancer to immune checkpoint inhibitors. J Immunother Cancer 9 , e001845 (2021). Martens, A. et al. Baseline peripheral blood biomarkers associated with clinical outcome of advanced melanoma patients treated with ipilimumab. Clin Cancer Res 22 , 2908 (2016). Koch, S. et al. Multiparameter flow cytometric analysis of CD4 and CD8 T cell subsets in young and old people. Immunity and Ageing 5 , (2008). Yu, X., Zhang, L., Chaudhry, A., Rapaport, A. S. & Ouyang, W. Unravelling the heterogeneity and dynamic relationships of tumor-infiltrating T cells by single-cell RNA sequencing analysis. J Leukoc Biol 107 , 917–932 (2020). Zheng, L. et al. Pan-cancer single-cell landscape of tumor-infiltrating T cells. Science (1979) 374 , (2021). Speiser, D. E. et al. Human CD8+ T cells expressing HLA-DR and CD28 show telomerase activity and are distinct from cytolytic effector T cells. Eur J Immunol 31 , 459–466 (2001). Miggelbrink, A. M. et al. CD4 T-Cell exhaustion: Does it exist and what are its roles in cancer? Clinical Cancer Research 27 , 5742–5752 (2021). Lingel, H. & Brunner-Weinzierl, M. C. CTLA-4 (CD152): A versatile receptor for immune-based therapy. Semin Immunol 42 , 101298 (2019). Brenchley, J. M. et al. Expression of CD57 defines replicative senescence and antigen-induced apoptotic death of CD8+ T cells. Blood 101 , 2711–2720 (2003). Verma, K. et al. Human CD8+ CD57- TEMRA cells: Too young to be called ‘old’. PLoS One 12 , e0177405 (2017). Lee, S. W. et al. CD8+ TILs in NSCLC differentiate into TEMRA via a bifurcated trajectory: deciphering immunogenicity of tumor antigens. J Immunother Cancer 9 , e002709 (2021). Bernal-Estévez, D., Sánchez, R., Tejada, R. E. & Parra-López, C. Chemotherapy and radiation therapy elicits tumor specific T cell responses in a breast cancer patient. BMC Cancer 16 , (2016). Kuznetsova, M. et al. Cytotoxic activity and memory T cell subset distribution of in vitro-stimulated CD8+ T cells specific for HER2/neu epitopes. Front Immunol 10 , (2019). Melchiotti, R., Gracio, F., Kordasti, S., Todd, A. K. & de Rinaldis, E. Cluster stability in the analysis of mass cytometry data. Cytometry Part A 91 , 73–84 (2017). Garcia, S. et al. Thoracic Oncology Clinical Trial Eligibility Criteria and Requirements Continue to Increase in Number and Complexity. Journal of Thoracic Oncology 12 , 1489–1495 (2017). Cooper, R. A., Chai, Y. & Nieva, J. Effect of sponsor on enrollment criteria in non-small cell lung cancer clinical trials. J Cancer Policy 33 , 100336 (2022). Additional Declarations No competing interests reported. Supplementary Files Supp.Table1.xlsx Supp.Table4.xlsx Supp.Table2.xlsx Supp.Table5.xlsx Supp.Table3.xlsx Graphicalabstract.jpg SupplementaryFigures.docx 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-8656976\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":true,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":578897253,\"identity\":\"d26e456c-6ee0-45fb-adee-995c7c0f8a82\",\"order_by\":0,\"name\":\"Laura Boyero\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzElEQVRIiWNgGAWjYBACPgkog02CsYHhAzFa2JC1MM4gSQsDkMHMQ5QW6eZjEh/32DDwSTc3f7Ztq2Pg5z9AQIvMsTTJGc/SgIyDbdK5bYcZJGckEHJYjpk0z4HD9W0SiW3MuW0HGAxuEPQLUMufA4eBjMTmz5ZAh9mfJ+QwkBYGiJYGacY2ZgYDBkIOkzmWbNlzAOIXyZ5zh3kkbhDQwi/dfPDGjwM2DPKz2x9/+FFWJ8ffT8BhQMAigcwjKmqYiUomo2AUjIJRMIIBAJfrOq1ZFFc2AAAAAElFTkSuQmCC\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Laura\",\"middleName\":\"\",\"lastName\":\"Boyero\",\"suffix\":\"\"},{\"id\":578897254,\"identity\":\"489056bf-9f68-4abc-bbeb-73ef43fd876d\",\"order_by\":1,\"name\":\"María G Velasco-Domínguez\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"María\",\"middleName\":\"G\",\"lastName\":\"Velasco-Domínguez\",\"suffix\":\"\"},{\"id\":578897255,\"identity\":\"d61afc4a-fc9b-4931-8915-412ecf883093\",\"order_by\":2,\"name\":\"Alejandro Castillo-Peña\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Alejandro\",\"middleName\":\"\",\"lastName\":\"Castillo-Peña\",\"suffix\":\"\"},{\"id\":578897256,\"identity\":\"d30a9164-9c7a-497d-b31f-80029067f40e\",\"order_by\":3,\"name\":\"Johana C Benedetti\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Medical Oncology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Johana\",\"middleName\":\"C\",\"lastName\":\"Benedetti\",\"suffix\":\"\"},{\"id\":578897257,\"identity\":\"8c119f40-abfb-437b-bb22-ff24efef40c5\",\"order_by\":4,\"name\":\"Sofía Boyero-Corral\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Sofía\",\"middleName\":\"\",\"lastName\":\"Boyero-Corral\",\"suffix\":\"\"},{\"id\":578897258,\"identity\":\"7ea824c4-e51b-4182-a3ad-bba6dd4566ec\",\"order_by\":5,\"name\":\"Miriam Alonso-García\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Medical Oncology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Miriam\",\"middleName\":\"\",\"lastName\":\"Alonso-García\",\"suffix\":\"\"},{\"id\":578897259,\"identity\":\"f3f03381-2b79-45a0-9cdf-2780b4339b7f\",\"order_by\":6,\"name\":\"Amparo Sánchez-Gastaldo\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Medical Oncology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Amparo\",\"middleName\":\"\",\"lastName\":\"Sánchez-Gastaldo\",\"suffix\":\"\"},{\"id\":578897260,\"identity\":\"bf8b3fb0-b2af-4f24-95ab-9d0ddf5a058e\",\"order_by\":7,\"name\":\"Yolanda M Pacheco\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Immunology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yolanda\",\"middleName\":\"M\",\"lastName\":\"Pacheco\",\"suffix\":\"\"},{\"id\":578897261,\"identity\":\"92ef9510-eafc-487b-ad28-d4dd27fb630e\",\"order_by\":8,\"name\":\"Sara Lopez-Martinez\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Immunology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Sara\",\"middleName\":\"\",\"lastName\":\"Lopez-Martinez\",\"suffix\":\"\"},{\"id\":578897262,\"identity\":\"79a52d61-c25c-4487-a92d-c3adff2eec4d\",\"order_by\":9,\"name\":\"Miguel A Muñoz-Fuentes\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Miguel\",\"middleName\":\"A\",\"lastName\":\"Muñoz-Fuentes\",\"suffix\":\"\"},{\"id\":578897263,\"identity\":\"42267b12-6ab1-40ac-b73c-6996f8a75092\",\"order_by\":10,\"name\":\"Sonia Molina-Pinelo\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Sonia\",\"middleName\":\"\",\"lastName\":\"Molina-Pinelo\",\"suffix\":\"\"},{\"id\":578897264,\"identity\":\"62b0c115-74e8-4a37-9e18-802b4c0e2a11\",\"order_by\":11,\"name\":\"Reyes Bernabé\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Department of Medical Oncology, Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital, CSIC, University of Seville\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Reyes\",\"middleName\":\"\",\"lastName\":\"Bernabé\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2026-01-21 08:16:25\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-8656976/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-8656976/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":101189026,\"identity\":\"60d7f9e2-8eb3-48a2-9390-fd094ff897ea\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":384979,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eA) Flowchart illustrating the patient selections process and inclusion criteria. B) Schematic representation of the sample collection process. C) Kaplan–Meier survival curves for PFS and OS in R and NR groups. The number of patients at risk is shown below the x-axis. Dashed lines indicate the median PFS and OS time points, defined as the time at which 50% of patients remained progression-free or alive, respectively. E) UMAP projection illustrating the distribution of the major immune cell types. F) Expression of lineage-defining markers used to identify major immune cell populations. G) Proportional bar chart illustrating the differences in the baseline frequency of major immune cell subsets between R and NR groups.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"6.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/9c052c247af773c1260a8869.png\"},{\"id\":101189022,\"identity\":\"c80a0721-c458-4ebf-87cf-0f8c72d5aa99\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":430132,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eA) UMAP projection illustrating the clustering of manually gated CD3⁺ T cell subsets. B) Proportional bar chart and C) UMAP plots illustrating the differences in the baseline frequency of the CD4⁺ T cell subsets between R and NR groups. D) Proportional bar chart and E) UMAP plots depicting the differences in the baseline frequency of distinct CD8⁺ T cell subsets across R and NR groups.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/a70683b335ccfbb7d18f9dd9.png\"},{\"id\":101189018,\"identity\":\"bd123123-f8a5-4429-acb7-f6c3ec4aa508\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"png\",\"order_by\":3,\"title\":\"Figure 3\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":576728,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eA) Characterization of baseline thymic output of R and NR (n=10). Median values and interquartile ranges (IQR) are shown. Asterisks denote outliers. TREC, T-cell receptor excision circles. B) Heatmap showing the expression profiles of 32 immune markers across the 26 T cell subpopulations identified by unsupervised clustering (PhenoGraph and FlowSOM) of the CD3⁺ T cell compartment. \\u0026nbsp;C) Volcano plot showing the differential abundance of CD3⁺ T cell subpopulations between R and NR. The x-axis represents the log2 fold change (log2FC), and the y-axis shows the –log₁₀ of the False Discovery Rate (FDR). Subpopulations with an –log₁₀(FDR) \\u0026gt; 1 were considered statistically significant and are highlighted in dark grey. \\u0026nbsp;D) Boxplots showing the frequency (%) of CD3⁺ T cell subsets that exhibited statistically significant differences between R and NR. \\u0026nbsp;E) Median fold change (FC) in the expression of activation and exhaustion/senescence markers across CD3⁺ T cell subpopulations comparing R and NR groups. Statistically significant differences are indicated by asterisks. Color gradient represents FC magnitude, from higher (red) to lower (blue) values.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/707f801188604c2453bce69e.png\"},{\"id\":101206892,\"identity\":\"0b07f65f-1106-419c-9281-62da102de0e2\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 09:56:56\",\"extension\":\"png\",\"order_by\":4,\"title\":\"Figure 4\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":465415,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eA) Representative flow cytometry plot showing the gating strategy used to identify EMRA T cells (CD45RA⁺CCR7⁻) within CD4⁺ and CD8⁺ T cell subsets (CD3⁺). CM, central memory T cells; Naïve, naïve T cells; EM, effector memory T cells; EMRA, terminally differentiated effector memory T cells. B) Frequencies (%) of CD4⁺ and CD8⁺ EMRA T cells in R and NR groups following PHA stimulation. Medians are indicated by solid lines. C–D) Histograms showing the percentages of CD25⁺, CD69⁺, IFN-γ⁺, and GZMB⁺ cells within CD4⁺ EMRA T cells (C) and CD8⁺ EMRA T cells (D) in R (blue) and NR (red) groups under PHA stimulation. Data are concatenated per group for visualization.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/6608c9c525a1054c13bd32b2.png\"},{\"id\":101189021,\"identity\":\"8eb209fe-05c9-40ab-acd4-2fa436efb38b\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"png\",\"order_by\":5,\"title\":\"Figure 5\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":453512,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eA) Variable importance plot from the Random Forest model identifying key predictors of immunotherapy response in NSCLC. The horizontal axis shows the mean decrease in Gini index, where higher values indicate greater variable importance. The vertical axis lists the features included in the model. B-D) Boxplots showing the expression levels of the three top-ranked parameters identified—CD25/CTLA-4 ratio (B), CTLA-4 expression (C), and HLA-DR expression (D) in CD4⁺ EMRA T cells—between R and NR groups. \\u0026nbsp;E) Kaplan–Meier survival curves illustrate differences in PFS and OS according to expression levels. Patients were stratified into high (\\u0026gt; median) and low (≤ median) expression groups. The number of patients at risk is shown below the x-axis.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"4.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/acd8f920937bd23b676f11ed.png\"},{\"id\":101207033,\"identity\":\"b8d6af95-6565-4f35-a9f0-1ef67458be8e\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 09:57:10\",\"extension\":\"png\",\"order_by\":6,\"title\":\"Figure 6\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":406006,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eCD4⁺ EMRA ImmunoPredict Score (CEIPS), which integrates the CD25/CTLA-4 ratio, CTLA-4 expression, and HLA-DR expression in CD4⁺ EMRA T cells, predicts pembrolizumab response. \\u0026nbsp;A) Receiver operating characteristic (ROC) curve evaluating the predictive performance of the CD4⁺ EMRA ImmunoPredict Score (CEIPS), which integrates the three CD4⁺ EMRA-associated parameters, in predicting response to pembrolizumab. \\u0026nbsp;B) Stacked bar plots showing the distribution of CEIPS scores among R and NR. \\u0026nbsp;C) PFS and D) OS according to CEIPS-based risk stratification (Low-Risk vs High-Risk). \\u0026nbsp;Results are presented as Kaplan–Meier survival curves and compared using the log-rank test; the number of patients at risk is shown below the x-axis. \\u0026nbsp;E) Swimmer plots of PFS and F) OS , stratified by CEIPS risk group. Event status is indicated as 1 = deceased, 0 = alive.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"5.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/727e056d166d71dcdb43cd0b.png\"},{\"id\":101208667,\"identity\":\"9c5a3fed-8d6d-44c0-ba77-952bc02a5486\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 10:10:41\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":3785963,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/006609c6-05cc-4876-abbc-29d7c4eee519.pdf\"},{\"id\":101189017,\"identity\":\"1fa104a3-04a3-48bd-ae70-cce41f4c58f9\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"xlsx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":12095,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Supp.Table1.xlsx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/28d927a11866c4b700690dd2.xlsx\"},{\"id\":101189019,\"identity\":\"6460d455-77cd-454f-b339-2d8f184caad7\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"xlsx\",\"order_by\":2,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":26073,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Supp.Table4.xlsx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/6b299df778028dd3ce35f969.xlsx\"},{\"id\":101206949,\"identity\":\"3b08e9f5-f9c5-4f95-9828-e88cd845b241\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 09:57:02\",\"extension\":\"xlsx\",\"order_by\":3,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":11231,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Supp.Table2.xlsx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/63ee473614019174f3d7bd9d.xlsx\"},{\"id\":101189028,\"identity\":\"7e975008-b95e-4122-add3-8b04caeeef22\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"xlsx\",\"order_by\":4,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":110458,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Supp.Table5.xlsx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/17872e1143ce0f23a786011e.xlsx\"},{\"id\":101189029,\"identity\":\"887acc89-8245-414c-9808-1bed9aef50ae\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"xlsx\",\"order_by\":5,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":13469,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Supp.Table3.xlsx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/f035ec31451827be81ddff39.xlsx\"},{\"id\":101189024,\"identity\":\"ddf76da5-b99b-4207-a126-1f32b247c680\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"jpg\",\"order_by\":6,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":341002,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Graphicalabstract.jpg\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/085b272741c052726ab649e6.jpg\"},{\"id\":101189030,\"identity\":\"131552e1-fa9c-4e7c-ace5-22bde566e30f\",\"added_by\":\"auto\",\"created_at\":\"2026-01-27 06:48:31\",\"extension\":\"docx\",\"order_by\":7,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":2514906,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"SupplementaryFigures.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-8656976/v1/52852b9c8cd274eef586a45a.docx\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"Single-Cell Immune Profiling and Machine Learning Reveal a Predictive Immune Signature for Immunotherapy Response in NSCLC\",\"fulltext\":[{\"header\":\"BACKGROUND\",\"content\":\"\\u003cp\\u003eImmune checkpoint inhibitors (ICIs) have revolutionized the treatment landscape for patients with metastatic non-small cell lung cancer (NSCLC) lacking oncogenic drivers since their approval in 2015, significantly improving both survival and quality of life. However, although ICIs are effective in a subset of patients, only approximately 20% of those with NSCLC achieve durable responses and long-term clinical benefit\\u003csup\\u003e1\\u003c/sup\\u003e. This limited efficacy contributes to NSCLC remaining the leading cause of cancer-related death worldwide\\u003csup\\u003e2\\u003c/sup\\u003e, despite a gradual decline in mortality rates over the years, largely due to low screening uptake and the fact that over half of patients are diagnosed at advanced stages\\u003csup\\u003e3\\u003c/sup\\u003e. This underscores the urgent need for reliable predictive biomarkers to guide treatment decisions. To date, extensive efforts have been made to identify such biomarkers, including programmed cell death protein 1 ligand (PD-L1) expression\\u003csup\\u003e4\\u003c/sup\\u003e, tumor mutational burden (TMB)\\u003csup\\u003e5\\u003c/sup\\u003e, and gut microbiota composition\\u003csup\\u003e6\\u003c/sup\\u003e, among others. However, none have demonstrated sufficient robustness or clinical utility in the context of NSCLC.\\u003c/p\\u003e\\n\\u003cp\\u003eICIs function by blocking the interaction between PD-1 (sustainably expressed on exhausted T cells) and its ligands PD-L1 and PD-L2 (commonly upregulated in the tumor microenvironment), thereby enhancing or restoring the immune response. This occurs through the promotion of cytotoxic effector activity\\u003csup\\u003e7–9\\u003c/sup\\u003e and/or by disrupting immune tolerance mechanisms against lung tumor cells\\u003csup\\u003e10\\u003c/sup\\u003e. Since ICIs primarily exert their effects by modulating T cell function, the baseline immunophenotypic landscape of peripheral blood mononuclear cells (PBMCs) may play a pivotal role in determining therapeutic response\\u003csup\\u003e11\\u003c/sup\\u003e, offering valuable insights into the systemic immune status of patients. In this context, advanced single-cell technologies such as mass cytometry enable high-dimensional profiling of peripheral blood, allowing in-depth functional characterization of immune populations, including states of activation, senescence, or exhaustion, that may reflect treatment responsiveness\\u003csup\\u003e12–14\\u003c/sup\\u003e. This approach may help identify peripheral biomarkers associated with clinical outcomes, thereby supporting personalized treatment strategies and reducing unnecessary exposure to ineffective therapies, with both clinical and economic implications\\u003csup\\u003e15\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003eIn this study, we conducted comprehensive PBMC profiling and investigated biomarkers associated with clinical outcomes by integrating mass cytometry with machine learning in a real-world, unselected cohort of metastatic NSCLC patients treated with first-line pembrolizumab. Additionally, we performed in vitro T cell stimulation assays to assess functional immune characteristics under controlled conditions. Based on these findings, we developed a scoring system based on a single functional T cell subpopulation, designed for direct application in clinical settings.\\u003c/p\\u003e\"},{\"header\":\"METHODS\",\"content\":\"\\u003ch2\\u003ePatients and study design\\u003c/h2\\u003e\\n\\u003cp\\u003ePatients were included through a search on the Virgen del Rocío Hospital (Seville, Spain) database for those diagnosed with NSCLC and who had received treatment with immunotherapy and with available follow-up data between the October 2017 and April 2021. Patient selection criteria included stage IV, having been treated with Pembrolizumab in first line, availability of a baseline sample and passing quality controls regarding viability of thawed PBMCs. A flowchart for patient selection and exclusion criteria is represented in Figure 1A. Out of a total of 145 available patients, we finally analyzed samples from those who met the eligibility criteria (n = 21).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003ePatients received intravenous single-agent pembrolizumab at a dose of 200 mg every 3 weeks (21 days). Elegilibility for pembrolizumab treatment required PD-L1 expression levels in tumour cells to be greater than or equal to 50% in all patients. Treatment response at 6 months was evaluated in all patients with analyzed samples, according to RECIST 1.1 criteria, based on computed tomography and assessed by radiologists and medical oncologists (Fig. 1B). The objective tumor response was defined as the best clinical response observed during the course of treatment. Patients were stratified based on response into a disease control rate (DCR) group—comprising stable disease (SD), partial response (PR), and complete response (CR)—and a progressive disease (PD) group. \\u0026nbsp;In this study, patients classified as having disease control rate (DCR) were considered responders (R), whereas those with progressive disease (PD) were considered non-responders (NR). All participants provided written informed consent prior to enrollment, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Coordinating Committee on Biomedical Research Ethics of Andalucía (0944-N-20).\\u003c/p\\u003e\\n\\u003ch2\\u003eSample processing\\u003c/h2\\u003e\\n\\u003cp\\u003ePeripheral blood samples were collected at baseline prior to pembrolizumab administration using two 8mL Cell Preparation Tubes (CPT) with sodium heparin (Vacutainer® CPT™ Mononuclear Cell Preparation Tubes, BD Biosciences). Samples were maintained at room temperature and processed within 3 hours of collection to ensure cell viability. Peripheral blood mononuclear cells (PBMC) were isolated by centrifugation following the manufacturer's instructions, washed with phosphate-buffered saline (PBS) and cryopreserved at -80°C in fetal calf serum (FCS) with 10% DMSO until processing. Quality control measures were implemented to ensure sample integrity, including viability assessment prior to cryopreservation and after thawing, before subsequent analyses.\\u003c/p\\u003e\\n\\u003ch2\\u003eMass cytometry\\u003c/h2\\u003e\\n\\u003cp\\u003eAfter thawing, 3 million PBMCs were incubated with 20 U/μL of DNase in RPMI-1640 media for 1 hour at 37ºC and 5% CO\\u003csub\\u003e2\\u003c/sub\\u003e, followed by a 2-hour resting period at 37ºC in the same supplemented culture medium to allow cell recovery. Samples were included only if cell viability exceeded 85%. PBMCs were then washed and incubated with 0.5mg/mL of FcR-blocking (Human TruStain FcX, Fluidigm) for 10 minutes at room temperature (RT)\\u0026nbsp;to prevent nonspecific binding.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eCells were subsequently stained for 30 minutes at RT using antibodies from the Maxpar Direct Immune Profiling Assay kit (Fluidigm) (Supp. Table 1), along with an additional pre-conjugated anti-CD279 (PD-1) antibody. Following surface staining, cells were permeabilized using Maxpar Perm-S Buffer (Fluidigm) and stained for 30 minutes at RT with an in-house-labeled intracellular anti-CTLA-4 antibody. All antibodies used for mass cytometry are listed in Supplementary Table 1. The in-house antibody labeling was performed using the Maxpar X8 Antibody Labeling Kit (Fluidigm), according to the manufacturer’s instructions.\\u003c/p\\u003e\\n\\u003cp\\u003eAfter washing, stained PBMCs were fixed incubating with 1.6% formaldehyde for 10 min. Cells were then incubated overnight with 125 nM of Cell-ID Intercalator-Ir solution (Fluidigm) for cell viability at 4ºC. Finally, cells were washed and stored at -80ºC until acquisition on a CyTOF2-Helios® instrument (Fluidigm) with a minimum of 200,000 events per sample.\\u003c/p\\u003e\\n\\u003cp\\u003eThis antibody panel enabled the identification of innate immune populations, including monocytes, natural killer (NK) cells, plasmacytoid and myeloid dendritic cells (DC), and granulocytes, as well as adaptive immune populations, such as naïve, central memory, effector memory, and terminally differentiated effector memory (EMRA) CD4⁺ and CD8⁺ T cells; CD4⁺CD8⁺ and CD4⁻CD8⁻ T cells; regulatory T cells (Tregs); γδ T cells; B cells; plasmablasts; and NKT cells. In addition, functional markers including CD127, CD25, CD38, CTLA-4, HLA-DR, PD-1, CD28, and CD57 were evaluated across all of the adaptative immune subsets.\\u003c/p\\u003e\\n\\u003ch2\\u003eMass Cytometry Data Analysis\\u003c/h2\\u003e\\n\\u003cp\\u003eThe resulting data were processed using a pipeline that included bead-based normalization, bead exclusion, cell doublet exclusion, and live/dead cell exclusion. To identify stratifying cell clusters based on selected markers from the mass cytometry data, Uniform Manifold Approximation and Projection (UMAP) was performed using FlowJo v10.10.0 using the DownSample v3.3.1 and UMAP v4.1.1 plug-ins with default settings. First, samples were downsampled to equal numbers of events and then concatenated for analysis. The PhenoGraph v4.0.5 plug-in was used to estimate the optimal number of clusters based on the expression of multiple parameters. Subsequently, the FlowSOM v4.1.0 plug-in was applied for clustering analysis. The UMAP algorithm was run using the marker expression specified in each figure and visualized as a heatmap and density plot.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eSingle-cell data has been clustered using the \\u003cem\\u003eFlowSOM\\u003c/em\\u003e R package\\u003csup\\u003e16\\u003c/sup\\u003e and labeled using the Ek'Balam algorithm\\u003csup\\u003e17\\u003c/sup\\u003e. Cell subset definitions follow Maecker et al.\\u003csup\\u003e18\\u003c/sup\\u003e and Finak et al.\\u003csup\\u003e19\\u003c/sup\\u003e. Cluster labeling, method implementation, and visualization were done through the Astrolabe Cytometry Platform (Astrolabe Diagnostics, Inc.) (Supp. Table 2). Differential abundance analysis was done using the \\u003cem\\u003eedgeR\\u003c/em\\u003e R package\\u003csup\\u003e20,21\\u003c/sup\\u003e following the method outlined in Lun et al.\\u003csup\\u003e22\\u003c/sup\\u003e. Differential expression analysis was done using the \\u003cem\\u003elimma\\u003c/em\\u003e R package\\u003csup\\u003e23\\u003c/sup\\u003e following the method outlined in Weber et al.\\u003csup\\u003e24\\u003c/sup\\u003e\\u003c/p\\u003e\\n\\u003ch2\\u003eThymic output quantification\\u003c/h2\\u003e\\n\\u003cp\\u003eThymic output was estimated based on the δ/β T-cell receptor excision circles (TRECs) ratio, measured by droplet digital PCR (ddPCR), modified from Ferrando-Martínez \\u003cem\\u003eet al.\\u003c/em\\u003e\\u003csup\\u003e25\\u003c/sup\\u003e to update the measure by droplet digital PCR (ddPCR). Briefly, genomic DNA was extracted from an average of 3×10⁶ remaining PBMCs using the QIAamp DNA Mini Kit (Qiagen), following the manufacturer’s instructions. Each ddPCR reaction, containing 150 ng of PBMC-derived DNA, was performed using highly sensitive specific probes as previously described\\u003csup\\u003e26\\u003c/sup\\u003e in a final volume of 20 μL using ddPCR Supermix for Probes (no dUTP, Bio-Rad) on a QX200 Droplet Digital PCR system (Bio-Rad). Data were analyzed using QuantaSoft software version 1.7.1 (Bio-Rad).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003ch2\\u003eAbsolute telomere length\\u003c/h2\\u003e\\n\\u003cp\\u003eCopy number quantification of the telomere repeat sequence and the single-copy gene 36B4 was performed by qPCR, following the protocol described by O’Callaghan \\u0026amp; Fenech (2011)\\u003csup\\u003e27\\u003c/sup\\u003e. Each reaction contained 20 ng of genomic DNA from remaining PBMCs. Primer sequences were as follows (5′→3′): Telomere Forward (CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGGTTTGGGTT) and\\u0026nbsp;Reverse (GGCTTGCCTTACCCTTACCCTTACCCTTACCCTTACCCT); 36B4 Forward (CAGCAAGTGGGAAGGTGTAATCC) and Reverse (CCCATTCTATCATCAACGGGTACAA).\\u003c/p\\u003e\\n\\u003cp\\u003eStandard curves were generated using synthetic oligonucleotides obtained from Invitrogen for telomere sequence (TTAGGG)14\\u0026nbsp;and 36B4 gene (CAGCAAGTGGGAAGGTGTAATCCGTCTCCACAGACAAGGCCAGGACTCGTTTGTACCCGTTGATGATAGAATGGG). To maintain a constant load of 20 ng per reaction, pUC19 plasmid DNA (Sigma‑Aldrich) was added to each standard tube. As a positive control, genomic DNA from the 1301 cell line (European Collection of Authenticated Cell Cultures, ECACC) was used. Reactions were run on a CFX Duet real-time PCR system (Bio-Rad) and absolute telomere length were expressed as kilobases of telomeric repeats per single-copy gene (kb/SCG), as previously described\\u003csup\\u003e27\\u003c/sup\\u003e.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003ch2\\u003eFlow cytometry\\u003c/h2\\u003e\\n\\u003cp\\u003ePBMCs were isolated as previously described, thawed, and incubated with Benzonase (Sigma-Aldrich) for 15 minutes at 37°C. PBMCs (5×10⁶ cells/mL) were then cultured in RPMI-1640 medium (Invitrogen Life Technologies) supplemented with human serum (Sigma-Aldrich) and antibiotics (Gibco), and stimulated with 5 μg/mL phytohemagglutinin (PHA, Thermo Fisher) for 24 hours at 37°C, in a 5% CO₂ atmosphere. To block cytokine secretion, GolgiStop (BD Biosciences) and Brefeldin A (BioLegend) were added to the cultures for 4 hours, following the manufacturers’ instructions.\\u003c/p\\u003e\\n\\u003cp\\u003eCells were washed and resuspended in DPBS (Gibco) at a concentration of 2×10⁶ cells/mL, and stained with Fixable Dead Cell Yellow (Invitrogen) at a 1:10,000 dilution for 30 minutes in the dark. Subsequently, cells were washed, fixed, and permeabilized using Cytofix/Cytoperm reagent (BD Biosciences) for intracellular staining, incubated for 20 minutes at 4°C in the dark. Finally, cells were stained overnight at 4°C in a wash buffer solution containing saponin, Brilliant Stain Buffer (BD Biosciences), and Fc Block (BD Biosciences), using pre-titrated fluorochrome-conjugated monoclonal antibodies: anti-CD3-Alexa700 (1:5000, Invitrogen), anti-CD4-APC-Cy7 (1:5000, BD Biosciences), anti-CD8-BV711 (1:100, Invitrogen), anti-CCR7-RB613 (1:5000, BD Biosciences), anti-CD45RA-PE-Cy7 (1:4000, Invitrogen), anti-CD25-BV605 (1:20, BD Biosciences), anti-CD69-APC (1:1000, BD Biosciences), anti-IFN-γ-BV421 (1:1000, BioLegend) and anti-GZMB-FITC (1:25, BD Biosciences). Data acquisition was performed using a Cytek Aurora spectral cytometer (Cytek Biosciences), and data were analyzed with FlowJo software (v10.8.1).\\u003c/p\\u003e\\n\\u003ch2\\u003eStatistical analyses\\u003c/h2\\u003e\\n\\u003cp\\u003eData are presented as medians with interquartile ranges (IQRs). Statistical comparisons between groups were conducted using the non-parametric Mann-Whitney U test for continuous variables. All statistical analyses and data visualizations were performed using IBM SPSS Statistics v26.0, GraphPad Prism v6.01 (GraphPad Software, Inc.), or the SRplot platform (http://www.bioinformatics.com.cn)\\u003csup\\u003e28\\u003c/sup\\u003e, employing non-parametric statistical tests in all cases. Outliers were identified and excluded using the ROUT method (Q = 1%). Survival analyses were performed by Kaplan-Meier method and comparisons between groups were performed using the log-rank test. Progression-free survival (PFS) was defined as the time from the initiation of pembrolizumab treatment to either clinical or radiographic disease progression, death from any cause, or censoring at the data cut-off date. Overall survival (OS) was defined as the time from the start of pembrolizumab treatment to death from any cause or censoring at the data cut-off. Survival curves were plotted using the survminer package in R, displaying 95% confidence intervals (CIs). A p-value \\u0026lt; 0.05 was considered statistically significant.\\u003c/p\\u003e\\n\\u003ch2\\u003eMachine Learning-Based Predictive Model\\u003c/h2\\u003e\\n\\u003cp\\u003eTo develop a predictive model of response to first-line pembrolizumab in stage IV NSCLC based on differential marker expression in T cell subpopulations, we employed Random Forest, a supervised machine learning classification algorithm, to identify the most informative variables. This analysis was performed using the caret package in R v4.3.2. Since no independent dataset was available, model performance was validated using bootstrapping, a widely accepted resampling technique that helps mitigate overfitting and assess model stability in small datasets. Variable importance was calculated based on the Gini impurity index.\\u003c/p\\u003e\\n\\u003cp\\u003eA simplified model was then constructed using the top-ranked variables. Model performance metrics included the area under the curve (AUC), sensitivity, specificity, precision, and Cohen’s kappa coefficient. The ROC curve was plotted using the ROCR package in R.\\u003c/p\\u003e\"},{\"header\":\"RESULTS\",\"content\":\"\\u003ch2\\u003ePatient Characteristics and Clinical Outcomes\\u003c/h2\\u003e\\n\\u003cp\\u003eBaseline clinicopathological and demographic characteristics of the patients are summarized in\\u0026nbsp;Table 1. A total of 21 subjects (aged 57 to 80 years) with stage IV NSCLC who received first-line pembrolizumab treatment were analyzed and categorized based on their treatment response at 6 months, according to RECIST 1.1 criteria. This resulted in n=10 (47.6%) responders (R) and n=11 (52.4%) non-responders (NR).\\u003c/p\\u003e\\n\\u003cp\\u003eThe median age of the patients was 67 years (range 57 to 80). The majority of patients were male (85.7%). Regarding performance status, 30.4% had an ECOG-PS score of 0, 47.8% had a score of 1, and 14.3% had a score of 2. The cohort was predominantly composed of patients with adenocarcinoma histology (81%). Regarding smoking history, over 70% of patients were either current smokers or had quit smoking within the last 10 years. The distribution patterns for age, sex, ECOG performance status, histological subtype, and smoking habits were comparable between the two treatment response groups, demonstrating no significant correlation between these characteristics and the response groups (p \\u0026gt; 0.05; Table 1). The median follow-up duration for the entire cohort was 25.23 months (95% CI 21.16-46.43). Patients in the R group had significantly better PFS and OS compared to those in the NR group (p \\u0026lt; 0.0001). Specifically, the median PFS and OS in the R group were 55.53 (95% CI 35.92-61.62) and 62.8 (95% CI 54.62-70.17) months, respectively, whereas in the NR group, the median PFS and OS were 3.5 (95% CI 2.07-4.93) and 13.9 months (95% CI 5.9-22.67) (Fig. 1C, D; Supp. Table 3).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eTable 1\\u003c/strong\\u003e. Baseline patient characteristics.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cimg width=\\\"567\\\" height=\\\"337\\\" src=\\\"data:image/emf;base64,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\\\" alt=\\\"image\\\"\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eECOG, Eastern Cooperative Oncology Group.\\u003c/p\\u003e\\n\\u003ch2\\u003eHigh-Dimensional Immune Profiling of Peripheral Blood\\u003c/h2\\u003e\\n\\u003cp\\u003eBaseline peripheral blood samples from patients with metastatic NSCLC treated with pembrolizumab were analyzed using mass cytometry to identify immune signatures associated with clinical response. To visualize and explore the high-dimensional single-cell data, we applied UMAP, a machine learning-based dimensionality reduction method, which revealed distinct clustering of immune cell subsets that differed between R and NR (Fig. 1E).\\u003c/p\\u003e\\n\\u003cp\\u003eWe identified the major immune cell types\\u0026mdash;including T lymphocytes (CD3\\u003csup\\u003e+\\u003c/sup\\u003e), B lymphocytes (CD19\\u003csup\\u003e+\\u003c/sup\\u003e), natural killer (NK) cells (CD56\\u003csup\\u003e+\\u003c/sup\\u003e), granulocytes (CD66b\\u003csup\\u003e+\\u003c/sup\\u003e), dendritic cells (DCs, CD11c\\u003csup\\u003e+\\u003c/sup\\u003e), and monocytes (CD14\\u003csup\\u003e+\\u003c/sup\\u003e)\\u0026mdash;through manual gating based on the expression of key phenotypic markers (Fig. 1F). We then compared the relative abundance of each major immune cell type between R and NR, and observed a trend toward a lower proportion of T and NK cells and a higher prevalence of monocytes, granulocytes, and dendritic cells (DCs) in R group compared to NR; however, these differences did not reach statistical significance (p \\u0026gt; 0.05) (Fig. 1G).\\u003c/p\\u003e\\n\\u003ch2\\u003eT Cell Subset Characterization and Clustering Analysis\\u003c/h2\\u003e\\n\\u003cp\\u003eGiven the observed difference in the relative abundance of CD3⁺ T cells between NR and R (fold change [FC] ~1.5), and considering the central role of T cells in the mechanism of action of immunotherapy, we sought to perform an in-depth characterization of T cell subpopulations to explore their potential as predictive biomarkers of treatment efficacy. To this end, we gated CD3⁺ T cells and generated a new UMAP to visualize the clustering patterns of T cell subsets (Fig. 2A; Supp. Fig. 1).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eWe then manually gated various T cell subsets, including CD4⁺ T cells (Fig. 2B, C), which encompassed na\\u0026iuml;ve (CD45RA⁺ CCR7⁺), central memory (CD45RA⁻ CCR7⁺), effector memory (CD45RA⁻ CCR7⁻), and EMRA cells\\u0026mdash;terminally differentiated effector memory cells re-expressing CD45RA (CD45RA⁺ CCR7⁻)\\u0026mdash;as well as regulatory T cells (Tregs; CD25⁺). Similarly, CD8⁺ T cells (Fig. 2D, E) were classified according to the same differentiation stages. In addition, we identified non-classical CD4/CD8 T cell subsets (Supp. Fig. 2A, B), comprising \\u0026gamma;\\u0026delta; T cells (TCR\\u0026gamma;\\u0026delta;⁺), double-positive T cells (CD4⁺ CD8⁺), and double-negative T cells (CD4⁻ CD8⁻).\\u003c/p\\u003e\\n\\u003cp\\u003eComparison of T cell subset frequencies revealed a higher proportion of CD4⁺ EMRA cells in the NR group compared to R (FC = 2.82), along with an increased frequency of CD8⁺ EMRA cells (FC = 1.24), indicating that both CD4⁺ and CD8⁺ EMRA subsets were enriched in NR patients (Fig. 2B, D). Conversely, the R group showed higher frequencies of CD4⁺ central memory T cells, CD8⁺ central memory T cells, and CD8⁺ effector memory T cells compared to NR (FC = 1.29, 1.50, and 1.46, respectively). No enrichment was observed between NR and R in the frequencies of na\\u0026iuml;ve T cells, regulatory T cells (Tregs), or non-classical CD4/CD8 subsets.\\u003c/p\\u003e\\n\\u003cp\\u003eGiven the compositional differences observed across T cell subsets, we next explored whether they might reflect underlying alterations in T cell ontogeny and maturation, processes that primarily occur in the thymus, and could indicate a weakened immune system in NR patients. To address this question, we quantified thymic output within the T cell compartment by comparing the \\u0026delta;/\\u0026beta; T-cell receptor excision circles (TRECs) ratio and absolute telomere length between R and NR patients, using a limited number of samples for which remaining PBMCs were available.\\u003c/p\\u003e\\n\\u003cp\\u003eThe NR group showed a lower thymic output than the R group, with median \\u0026delta;/\\u0026beta; TREC ratios of 0.4706 and 1.567, respectively (Fig. 3A). Although this difference did not reach statistical significance (p = 0.151), likely due to the limited sample size, exclusion of a single outlier identified by the ROUT method (Q = 1%) revealed a significant difference (p = 0.016). These findings suggest that R patients may retain greater thymic function and benefit from a broader and more diverse TCR repertoire, potentially contributing to their improved immune response. In contrast, the baseline for the telomere absolute length in circulating mononuclear cells showed no significant differences between response groups (Supp. Fig. 2C).\\u003c/p\\u003e\\n\\u003cp\\u003eTo complement and validate these findings through an unbiased approach, we performed unsupervised clustering on the CD3⁺ T cell population using the PhenoGraph and FlowSOM algorithms. This analysis, followed by manual annotation, allowed the identification of 26 distinct T cell subpopulations based on similarities in the expression profiles of multiple immune markers. Among these clusters, we identified all major T cell subsets, including CD4⁺ and CD8⁺ na\\u0026iuml;ve, central memory, effector memory, and EMRA cells, as well as Tregs, double-positive (CD4⁺CD8⁺), double-negative (CD4⁻CD8⁻), and \\u0026gamma;\\u0026delta; T cells (Fig. 3B; Supp. Table 4). Additionally, several subpopulations exhibited expression of activation, exhaustion and senescence markers (e.g., CD38, HLA-DR and CD57).\\u003c/p\\u003e\\n\\u003cp\\u003eA qualitative comparison of the identified T cell subpopulations between R and NR groups revealed higher frequencies of CD4⁺ EMRA, CD8⁺ EMRA, and CD4⁺CD8⁺ T cells in the NR group (FC = 3.66, 2.07, and 2.10, respectively).\\u003c/p\\u003e\\n\\u003cp\\u003eIn parallel with manual gating and qualitative analysis of CD3⁺ T cell subpopulations, an automated de-multiplexed algorithm was applied using the Astrolabe platform (Astrolabe Diagnostics Inc., Arlington, VA, USA). This algorithm implements a standardized analysis pipeline to quantitatively assess pre-defined immune cell subsets and identify statistically significant differences between groups. For this analysis, only CD3⁺ T cell populations were included in the automated workflow to compare R and NR (Fig. 3C, Supp. Fig. 2D).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eSignificant differences in the EMRA T cell compartment between R and NR were identified based on the output of the Astrolabe analysis.\\u003c/strong\\u003e Specifically, CD4⁺ EMRA and CD8⁺ EMRA subsets were found to be significantly increased in NR, with Log\\u003csub\\u003e2\\u003c/sub\\u003eFC of -1.94 and -1.38, and corresponding -Log₁₀ (FDR) values of 1.75 and 1.01, respectively (Fig. 3D). These results reinforce the enrichment of terminally differentiated T cell subsets in NR patients, particularly within the EMRA compartment, as revealed consistently by both manual and automated analyses.\\u003c/p\\u003e\\n\\u003ch2\\u003eFunctional Profiling of EMRA T Cells\\u003c/h2\\u003e\\n\\u003cp\\u003eWe also performed an in-depth analysis of baseline differential expression patterns of activation/regulatory and exhaustion/senescence markers across CD3⁺ T cell subpopulations, comparing R and NR using the standardized pipeline of the Astrolabe platform. Specifically, we evaluated the median FC in the expression of CD127, CD25, CD38, CTLA-4 and HLA-DR as activation/regulatory markers, and PD-1, CD28, and CD57 as exhaustion/senescence-associated markers. While no significant differences were observed for exhaustion/senescence markers, the R group exhibited significantly higher expression of CTLA-4 and HLA-DR in several CD3⁺ T cell subsets, with CD4⁺ EMRA cells showing the most pronounced differences (p = 0.007 and 0.002, respectively) (Fig. 3E; Supp. Table 5).\\u003c/p\\u003e\\n\\u003cp\\u003eGiven these findings, we sought to assess the actual functional activation capacity of the EMRA T cell compartment in R and NR patients. To this end, we conducted an in vitro stimulation assay in a limited subset of patients with available PBMCs (n = 8; 4 R and 4 NR), stimulated ex vivo with phytohemagglutinin (PHA), a potent T cell activator. Spectral flow cytometry was used to identify CD4⁺ and CD8⁺ EMRA T cells (CD45RA⁺CCR7⁻) (Fig. 4A). This analysis confirmed a trend toward a higher frequency of EMRA T cells in NR patients, with median values of 4.99% (NR) vs. 3.17% (R) for CD4⁺ EMRA T cells and 32.65% (NR) vs. 25.90% (R) for CD8⁺ EMRA T cells (Fig. 4B), although these differences did not reach statistical significance.\\u003c/p\\u003e\\n\\u003cp\\u003eFurther analysis of effector molecules and activation markers, including CD25, CD69, IFN-\\u0026gamma;, and granzyme B (GZMB), revealed higher expression levels in both CD4⁺ (Fig. 4C, Supp. Fig. 3A) and CD8⁺ EMRA T cells (Fig. 4D, Supp. Fig. 3B) from R patients compared to NR. These results suggest that R patients harbor a smaller EMRA T cell compartment, yet with greater functional capacity, than NR patients.\\u003c/p\\u003e\\n\\u003ch2\\u003eDevelopment of a Predictive Model for Pembrolizumab Response\\u003c/h2\\u003e\\n\\u003cp\\u003eWe sought to develop a robust and clinically relevant model to predict pembrolizumab efficacy in NSCLC. To minimize complexity and enhance translational applicability, we preselected variables that showed significant differences between R and NR groups; specifically, the relative abundance of CD4⁺ EMRA and CD8⁺ EMRA T cell subsets, along with the expression levels of key activation/regulatory markers (CD25, CTLA-4, and HLA-DR) within these populations.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eGiven the relevance of the functional balance between activation and inhibition, we also included biologically meaningful expression ratios, namely CD25/CTLA-4 and HLA-DR/CTLA-4, calculated within both CD4⁺ and CD8⁺ EMRA subsets. These ratios were hypothesized to reflect shifts in T cell functional states with potential predictive value.\\u003c/p\\u003e\\n\\u003cp\\u003eTo identify the most informative predictors and further reduce the variable set for the final model, we applied the non-parametric machine learning algorithm Random Forest to rank the contribution of each feature to classification performance (Fig. 5A). In addition to immunological features, we also included relevant clinical variables, such as age, sex, ECOG performance status, histology and smoking status, in the initial model. However, the relative contribution of these clinical variables was found to be \\u0026le;5%, and they were therefore excluded from the final model (Supp. Fig. 4).\\u003c/p\\u003e\\n\\u003cp\\u003eThe Random Forest analysis identified the CD25/CTLA-4 ratio in CD4⁺ EMRA T cells, along with the expression levels of CTLA-4 and HLA-DR within the same subset, as the top contributors to the predictive model. We independently evaluated the predictive and prognostic value of the three top-ranked parameters identified. All three variables (CD25/CTLA-4 ratio, CTLA-4 expression, and HLA-DR expression in CD4⁺ EMRA T cells) showed statistically significant differences between R and NR groups, with higher values observed in responders (Fig. 5B, C, D). Among them, the strongest association with response was observed for the CD25/CTLA-4 ratio (p \\u0026lt; 0.0001), followed by HLA-DR expression (p = 0.006) and CTLA-4 expression (p = 0.010). Kaplan-Meier survival analysis further demonstrated that elevated levels of all three parameters were significantly associated with improved PFS (p = 0.001 for CD25/CTLA-4 ratio, p = 0.002 for HLA-DR, and p = 0.023 for CTLA-4). In terms of OS, high levels of the CD25/CTLA-4 ratio (p = 0.007) and HLA-DR expression (p = 0.022) were associated with better outcomes, whereas CTLA-4 expression showed a similar trend but did not reach statistical significance (p = 0.091) (Fig. 5E).\\u003c/p\\u003e\\n\\u003cp\\u003eAlthough each of these parameters independently demonstrated predictive value for treatment response and survival, we investigated whether their combination could enhance overall predictive performance. Given the limited sample size, we employed bootstrap cross-validation within our cohort. This approach involved 100 random resampling iterations from the original dataset to minimize overfitting bias. The combined model, integrating the three CD4⁺ EMRA-associated parameters, effectively predicted response, achieving an area under the curve (AUC) of 0.908, with a sensitivity of 0.832 and a specificity of 0.869 (Fig. 6A). These findings support the robustness of the model and indicate that the observed results are not driven by a small number of outliers.\\u003c/p\\u003e\\n\\u003ch2\\u003eCD4⁺ EMRA ImmunoPredict Score (CEIPS) Stratifies Clinical Outcomes\\u003c/h2\\u003e\\n\\u003cp\\u003eTo facilitate future clinical translation of the predictive model, we developed a simplified scoring system, termed the \\u003cstrong\\u003eCD4⁺ EMRA ImmunoPredict Score (CEIPS)\\u003c/strong\\u003e. One point was assigned for each of the three model variables (CD25/CTLA-4 ratio, CTLA-4 expression, and HLA-DR expression in CD4⁺ EMRA T cells) when their values fell below the corresponding median, resulting in a cumulative score ranging from 0 to 3. Higher scores in this CEIPS were associated with poorer prognosis.\\u003c/p\\u003e\\n\\u003cp\\u003eAll patients with a score of 0 (S0) were classified within the R group, while scores of 3 (S3) were exclusively observed in the NR group. In addition, scores of 1 (S1) were more frequent in R, whereas scores of 2 (S2) predominated in NR (Fig. 6B). Based on this distribution, scores of 0 or 1 were categorized as Low Risk, and scores of 2 or 3 as High Risk. Kaplan-Meier analysis revealed significantly poorer PFS and OS in the High-Risk group compared to the Low-Risk group (p = 0.005 and p \\u0026lt; 0.001, respectively; Fig. 6C, D). These findings are further illustrated in Figures 6E and 6F, which show the individual PFS and OS outcomes for patients stratified by the CEIPS classification.\\u003c/p\\u003e\"},{\"header\":\"DISCUSSION\",\"content\":\"\\u003cp\\u003eWe analyzed the comprehensive PBMC profile of real-world metastatic NSCLC patients treated with first-line pembrolizumab, with treatment response assessed at 6 months and a median follow-up of 25.23 months. By combining single-cell mass cytometry, clustering and machine learning, and functional immune characterization, we developed a simple predictive model (CEIPS) based on a single functional T cell subpopulation. This integrative approach captures complementary immunological features, including early activation (CD25), sustained activation (HLA-DR), and immune regulation (CTLA-4), within the CD4⁺ EMRA subset, thereby providing a multidimensional perspective on T cell functionality relevant to immunotherapy response. The resulting score is clinically feasible and readily translatable, enabling real-time monitoring of ICI treatment response through noninvasive liquid-biopsy analysis.\\u003c/p\\u003e\\n\\u003cp\\u003eDespite the clinical success of ICIs across several tumor types\\u003csup\\u003e29\\u003c/sup\\u003e, including NSCLC\\u003csup\\u003e30,31\\u003c/sup\\u003e, only a minority of patients (~20%)\\u003csup\\u003e32\\u003c/sup\\u003e derive durable benefit. As clinical indications continue to expand, an increase in the proportion of non-responders is expected. Proposed biomarkers such as PD-L1 expression, microsatellite instability-high (MSI-H) status, and tumor mutational burden (TMB) have shown limited predictive efficacy\\u003csup\\u003e\\u0026nbsp;33\\u003c/sup\\u003e, underscoring the need for functional immune readouts that more accurately reflect host-tumor interactions. In this context, single-cell and machine learning approaches have enabled deeper dissection of immune heterogeneity, yet their clinical translation remains limited. Blood-based biomarkers, however, offer a readily accessible source of longitudinal information with the potential to complement or even surpass tumor-based biomarkers. In line with this landscape, our study identifies an immune signature with predictive value in NSCLC with immune checkpoint inhibitors, providing a liquid-biopsy-based framework for treatment monitoring.\\u003c/p\\u003e\\n\\u003cp\\u003eOur analysis of peripheral immune cell subpopulations revealed a lower proportion of CD3⁺ T lymphocytes in responders compared with non-responders at baseline.Notably, similar findings have been reported in melanoma\\u003csup\\u003e34\\u003c/sup\\u003e. Likewise, we observed an association between the circulating EMRA T cell compartment (CCR7⁻ CD45RA⁺) and treatment responsiveness, with responders displaying lower baseline frequencies of both CD4⁺ EMRA and CD8⁺ EMRA subsets. Independent data from melanoma\\u003csup\\u003e35,36\\u003c/sup\\u003e and mesothelioma\\u003csup\\u003e37\\u003c/sup\\u003e support these observations.Although seemingly counterintuitive, one plausible explanation is that responders harbor T cells with enhanced migratory capacity toward the tumor, where they infiltrate and exert antitumor activity\\u003csup\\u003e34\\u003c/sup\\u003e\\u003cstrong\\u003e.\\u0026nbsp;\\u003c/strong\\u003eThis redistribution of T cells from blood to tumor increases the number of tumor-infiltrating lymphocytes (TILs) in responders\\u003csup\\u003e38,39\\u003c/sup\\u003e,\\u0026nbsp;a hallmark of “hot” tumors\\u003csup\\u003e40\\u003c/sup\\u003e, and may indirectly contribute to the elevated frequency of circulating myeloid cells consistently reported in previous studies\\u003csup\\u003e34,41–43\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003eSimilarly, the peripheral increase in CD4⁺ and CD8⁺ EMRA clusters (T cells representing the most differentiated stage\\u003csup\\u003e44\\u003c/sup\\u003e) observed in non-responders may be attributable to differences in trafficking to the tumor, where they constitute only a small fraction of tumor-infiltrating T cells\\u003csup\\u003e45\\u003c/sup\\u003e. Indeed, EMRA T cell clusters exhibit the highest migration indices between blood and tumor tissues across most cancer types, and increased infiltration of CD8⁺ EMRA cells has been reported in some tumor samples from lung cancer and melanoma, and more consistently in renal cancer\\u003csup\\u003e46\\u003c/sup\\u003e, highlighting variability among cancer types. Beyond migration, EMRA clusters may also act as direct antitumor mediators. Supporting this notion, although EMRA T cells were less frequent in responders, they displayed greater in vitro functional activity, with higher expression of activation markers (CD25 and CD69) as well as effector molecules (IFN-γ and GZMB). Moreover, they showed increased expression of HLA-DR, a molecule associated with T cell activation and proliferation\\u003csup\\u003e47\\u003c/sup\\u003e, together with elevated CTLA-4 expression, typically linked to T cell exhaustion\\u003csup\\u003e48\\u003c/sup\\u003e in chronic conditions such as cancer, but also a key regulator of adaptive immunity\\u003csup\\u003e49\\u003c/sup\\u003e. This pattern is consistent with activation-induced inhibitory feedback and may reflect a highly activated but regulated T cell state, which may support initial response while potentially predisposing to immune exhaustion and resistance to anti-PD-1 monotherapy, thereby providing a rationale for combined PD-1/CTLA-4 blockade. Collectively, these data suggest that EMRA T cells from responders possess enhanced antitumor potential, dependent on their expressed gene signature. On this basis, we incorporated into the development of the CEIPS model not only the abundance of EMRA subsets but also their expression levels of key activation and regulatory markers. It is worth noting that the EMRA T cell pool\\u0026nbsp;increases\\u0026nbsp;with age and has been associated with immunosenescence\\u003csup\\u003e44\\u003c/sup\\u003e.Moreover, EMRA cells can be subdivided according to CD57 expression, with functional and proliferative (CD57⁻) subsets and senescent (CD57⁺) subsets, the latter characterized by increased telomere shortening and reduced TREC content\\u003csup\\u003e50–52\\u003c/sup\\u003e\\u003cstrong\\u003e.\\u0026nbsp;\\u003c/strong\\u003eIn our cohort, no significant age differences were found between responders and non-responders, suggesting that the differences observed in EMRA abundance may instead be attributable to reduced thymic output in non-responders, reflected in low δ/β TREC levels. This could limit the generation of new, potentially tumor-reactive TCRs, thereby favoring compensatory clonal expansion of preexisting populations in non-responders. Nevertheless, further studies will be needed to clarify this issue. Whether these clonotypes are specific for tumor antigens remains controversial, although supportive evidence has been reported in circulating EMRA cells from breast cancer patients\\u003csup\\u003e53,54\\u003c/sup\\u003e.\\u003c/p\\u003e\\n\\u003cp\\u003eThis study faced limitations typical of mass cytometry, such as a reduced cohort size due to the high cost of the technique and the challenge of high-dimensional data analysis\\u003csup\\u003e55\\u003c/sup\\u003e. The dimensionality reduction algorithm UMAP provides a qualitative overview but does not yield statistical analyses. Meanwhile, the Astrolabe pipeline streamlines data processing but offers limited standardization with relatively rigid analyses. Nevertheless, this automation helps to minimize operator bias and subjectivity. To mitigate these issues, supervised manual gating and statistical analyses were also implemented, along with functional assays conducted by experienced scientists to enable a deeper exploration of the data. On the other hand, this study is strengthened by the use of a real-world cohort of patients not selected by strict and complex enrollment criteria typically applied in clinical trials\\u003csup\\u003e56\\u003c/sup\\u003e, which often limit generalizability\\u003csup\\u003e57\\u003c/sup\\u003e. Furthermore, it adopts a translational approach by developing a clinically applicable score, simplified to a few parameters within a single T cell population, which can be monitored using small flow cytometry panels. Validation performed through machine learning and bootstrapping further supports the robustness of the model.\\u003c/p\\u003e\"},{\"header\":\"CONCLUSIONS\",\"content\":\"\\u003cp\\u003eIn conclusion, we identified circulating CD4⁺ and CD8⁺ EMRA T cells as predictors of response to first-line pembrolizumab therapy in metastatic NSCLC. We further developed the CD4⁺ EMRA ImmunoPredict Score (CEIPS), a predictive model based on activation and regulatory markers of CD4⁺ EMRA T cells that showed strong predictive capacity. These findings provide a noninvasive and feasible tool that could aid therapeutic decision-making in the context of personalized medicine and may also contribute to disease monitoring. Nonetheless, validation in larger prospective cohorts and across other tumor types will be required. If confirmed, this strategy could pave the way toward clinically implementable immune-functional biomarkers that complement or surpass current tumor-based assays.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cbr\\u003e\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003ch2\\u003eAvailability of data and materials\\u003c/h2\\u003e\\n\\u003cp\\u003eData are available in a public, open access repository. Data supporting this publication are available at ImmPort (https://www.immport.org) under study accession SDY3250.\\u003c/p\\u003e\\n\\u003ch2\\u003eEthics approval and consent to participate\\u003c/h2\\u003e\\n\\u003cp\\u003eAll participants provided written informed consent prior to enrollment, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Coordinating Committee on Biomedical Research Ethics of Andaluc\\u0026iacute;a (0944-N-20).\\u003c/p\\u003e\\n\\u003ch2\\u003eContributions\\u003c/h2\\u003e\\n\\u003cp\\u003eL.B. conceived the study, provided conceptualization, analyzed and discussed the results, and wrote the manuscript. M.G.V.-D., S.B.-C., and S.L.-M. performed and validated the experimental data. M.G.V.-D. and A.C.-P. conducted the statistical analyses. A.C.-P., L.B., and M.A.M.-F. developed the predictive model. J.C.B., M.A.-G., A.S.-G., and R.B. were responsible for patient recruitment, and J.C.B. evaluated clinical responses. S.M.-P., Y.M.P., and R.B. provided supervision, scientific expertise, and critical feedback. S.M.-P. and R.B. participated in funding acquisition. All authors reviewed and approved the final manuscript.\\u003c/p\\u003e\\u003cp\\u003e\\u003cstrong\\u003eFunding information\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe authors declare financial support was received for the research and/or publication of this article. LB and MAG were funded by Regional Ministry of Health and Consume of Andalucía (PI-0196-2025). AC-P was supported by an FPU22/04225 fellowship funded by the Spanish Ministry of Education. YMP\\u0026nbsp;was supported by grant CNS2023-144725 funded by AEI.\\u0026nbsp;SM-P was funded by the Ministry of Health and Social Welfare of Junta de Andalucia (Nicolás Monardes Program RC1-0005-2025), ISCIII (PI23/01679) and co-funded by FEDER from Regional Development European Funds (European Union).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eAcknowledgments\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eWe would like to acknowledge patients and their families for donating biological samples.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eCompeting interests\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe authors declare that they have no competing interests.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n\\u003cli\\u003eTopalian, S. L., Taube, J. M., Anders, R. A. \\u0026amp; Pardoll, D. M. Mechanism-driven biomarkers to guide immune checkpoint blockade in cancer therapy. \\u003cem\\u003eNat Rev Cancer\\u003c/em\\u003e \\u003cstrong\\u003e16\\u003c/strong\\u003e, 275 (2016).\\u003c/li\\u003e\\n\\u003cli\\u003eSiegel, R. L., Kratzer, T. B., Giaquinto, A. N., Sung, H. \\u0026amp; Jemal, A. Cancer statistics, 2025. \\u003cem\\u003eCa\\u003c/em\\u003e \\u003cstrong\\u003e75\\u003c/strong\\u003e, 10 (2025).\\u003c/li\\u003e\\n\\u003cli\\u003eOlivares-Hern\\u0026aacute;ndez, A. \\u003cem\\u003eet al.\\u003c/em\\u003e Immune checkpoint inhibitors in non-small cell lung cancer: from current perspectives to future treatments\\u0026mdash;a systematic review. \\u003cem\\u003eAnn Transl Med\\u003c/em\\u003e \\u003cstrong\\u003e11\\u003c/strong\\u003e, 354\\u0026ndash;354 (2023).\\u003c/li\\u003e\\n\\u003cli\\u003eKluger, H. M. \\u003cem\\u003eet al.\\u003c/em\\u003e PD-L1 Studies Across Tumor Types, its Differential Expression and Predictive Value in Patients Treated with Immune Checkpoint Inhibitors. \\u003cem\\u003eClin Cancer Res\\u003c/em\\u003e \\u003cstrong\\u003e23\\u003c/strong\\u003e, 4270 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003ePrasad, V. \\u0026amp; Addeo, A. The FDA approval of pembrolizumab for patients with TMB \\u0026gt;10 mut/Mb: was it a wise decision? No. \\u003cem\\u003eAnnals of Oncology\\u003c/em\\u003e \\u003cstrong\\u003e31\\u003c/strong\\u003e, 1112\\u0026ndash;1114 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eGrenda, A. \\u003cem\\u003eet al.\\u003c/em\\u003e Gut microbial predictors of first-line immunotherapy efficacy in advanced NSCLC patients. \\u003cem\\u003eScientific Reports 2025 15:1\\u003c/em\\u003e \\u003cstrong\\u003e15\\u003c/strong\\u003e, 1\\u0026ndash;12 (2025).\\u003c/li\\u003e\\n\\u003cli\\u003eZhou, Q. \\u003cem\\u003eet al.\\u003c/em\\u003e Blockade of Programmed Death-1 Pathway Rescues the Effector Function of Tumor-Infiltrating T Cells and Enhances the Antitumor Efficacy of Lentivector Immunization. \\u003cem\\u003eThe Journal of Immunology\\u003c/em\\u003e \\u003cstrong\\u003e185\\u003c/strong\\u003e, 5082\\u0026ndash;5092 (2010).\\u003c/li\\u003e\\n\\u003cli\\u003eBlank, C. \\u003cem\\u003eet al.\\u003c/em\\u003e Blockade of PD-L1 (B7-H1) augments human tumor-specific T cell responses in vitro. \\u003cem\\u003eInt J Cancer\\u003c/em\\u003e \\u003cstrong\\u003e119\\u003c/strong\\u003e, 317\\u0026ndash;327 (2006).\\u003c/li\\u003e\\n\\u003cli\\u003eWong, R. M. \\u003cem\\u003eet al.\\u003c/em\\u003e Programmed death-1 blockade enhances expansion and functional capacity of human melanoma antigen-specific CTLs. \\u003cem\\u003eInt Immunol\\u003c/em\\u003e \\u003cstrong\\u003e19\\u003c/strong\\u003e, 1223\\u0026ndash;1234 (2007).\\u003c/li\\u003e\\n\\u003cli\\u003eShepherd, F. A., Douillard, J. Y. \\u0026amp; Blumenschein, G. R. Immunotherapy for Non-small Cell Lung Cancer: Novel Approaches to Improve Patient Outcome. \\u003cem\\u003eJournal of Thoracic Oncology\\u003c/em\\u003e \\u003cstrong\\u003e6\\u003c/strong\\u003e, 1763\\u0026ndash;1773 (2011).\\u003c/li\\u003e\\n\\u003cli\\u003eRochigneux, P. \\u003cem\\u003eet al.\\u003c/em\\u003e Medical Treatment of Lung Cancer: Can Immune Cells Predict the Response? A Systematic Review. \\u003cem\\u003eFront Immunol\\u003c/em\\u003e \\u003cstrong\\u003e11\\u003c/strong\\u003e, 1036 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eCaligola, S., De Sanctis, F., Can\\u0026egrave;, S. \\u0026amp; Ugel, S. Breaking the Immune Complexity of the Tumor Microenvironment Using Single-Cell Technologies. \\u003cem\\u003eFront Genet\\u003c/em\\u003e \\u003cstrong\\u003e13\\u003c/strong\\u003e, (2022).\\u003c/li\\u003e\\n\\u003cli\\u003eChristodoulou, M. I. \\u0026amp; Zaravinos, A. Single-Cell Analysis in Immuno-Oncology. \\u003cem\\u003eInt J Mol Sci\\u003c/em\\u003e \\u003cstrong\\u003e24\\u003c/strong\\u003e, (2023).\\u003c/li\\u003e\\n\\u003cli\\u003eNikanjam, M., Kato, S. \\u0026amp; Kurzrock, R. Liquid biopsy: current technology and clinical applications. \\u003cem\\u003eJ Hematol Oncol\\u003c/em\\u003e \\u003cstrong\\u003e15\\u003c/strong\\u003e, 131 (2022).\\u003c/li\\u003e\\n\\u003cli\\u003eSeliger, B. \\u0026amp; Massa, C. CyTOF as a suitable tool for stratification and monitoring of cancer patients. \\u003cem\\u003eJ Transl Med\\u003c/em\\u003e \\u003cstrong\\u003e23\\u003c/strong\\u003e, 724 (2025).\\u003c/li\\u003e\\n\\u003cli\\u003eVan Gassen, S. \\u003cem\\u003eet al.\\u003c/em\\u003e FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data. \\u003cem\\u003eCytometry Part A\\u003c/em\\u003e \\u003cstrong\\u003e87\\u003c/strong\\u003e, 636\\u0026ndash;645 (2015).\\u003c/li\\u003e\\n\\u003cli\\u003eAmir, E. ad D. \\u003cem\\u003eet al.\\u003c/em\\u003e Development of a comprehensive antibody staining database using a standardized analytics pipeline. \\u003cem\\u003eFront Immunol\\u003c/em\\u003e \\u003cstrong\\u003e10\\u003c/strong\\u003e, 458295 (2019).\\u003c/li\\u003e\\n\\u003cli\\u003eMaecker, H. T., McCoy, J. P. \\u0026amp; Nussenblatt, R. Standardizing immunophenotyping for the Human Immunology Project. \\u003cem\\u003eNat Rev Immunol\\u003c/em\\u003e \\u003cstrong\\u003e12\\u003c/strong\\u003e, 191\\u0026ndash;200 (2012).\\u003c/li\\u003e\\n\\u003cli\\u003eFinak, G. \\u003cem\\u003eet al.\\u003c/em\\u003e Standardizing Flow Cytometry Immunophenotyping Analysis from the Human ImmunoPhenotyping Consortium. \\u003cem\\u003eSci Rep\\u003c/em\\u003e \\u003cstrong\\u003e6\\u003c/strong\\u003e, (2016).\\u003c/li\\u003e\\n\\u003cli\\u003eRobinson, M. D., McCarthy, D. J. \\u0026amp; Smyth, G. K. edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. \\u003cem\\u003eBioinformatics\\u003c/em\\u003e \\u003cstrong\\u003e26\\u003c/strong\\u003e, 139\\u0026ndash;140 (2010).\\u003c/li\\u003e\\n\\u003cli\\u003eMcCarthy, D. J., Chen, Y. \\u0026amp; Smyth, G. K. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. \\u003cem\\u003eNucleic Acids Res\\u003c/em\\u003e \\u003cstrong\\u003e40\\u003c/strong\\u003e, 4288\\u0026ndash;4297 (2012).\\u003c/li\\u003e\\n\\u003cli\\u003eLun, A. T. L., Richard, A. C. \\u0026amp; Marioni, J. C. Testing for differential abundance in mass cytometry data. \\u003cem\\u003eNat Methods\\u003c/em\\u003e \\u003cstrong\\u003e14\\u003c/strong\\u003e, 707\\u0026ndash;709 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003eRitchie, M. E. \\u003cem\\u003eet al.\\u003c/em\\u003e Limma powers differential expression analyses for RNA-sequencing and microarray studies. \\u003cem\\u003eNucleic Acids Res\\u003c/em\\u003e \\u003cstrong\\u003e43\\u003c/strong\\u003e, e47 (2015).\\u003c/li\\u003e\\n\\u003cli\\u003eWeber, L. M., Nowicka, M., Soneson, C. \\u0026amp; Robinson, M. D. diffcyt: Differential discovery in high-dimensional cytometry via high-resolution clustering. \\u003cem\\u003eCommun Biol\\u003c/em\\u003e \\u003cstrong\\u003e2\\u003c/strong\\u003e, (2019).\\u003c/li\\u003e\\n\\u003cli\\u003eFerrando-Mart\\u0026iacute;nez, S. \\u003cem\\u003eet al.\\u003c/em\\u003e A reliable and simplified sj/\\u0026beta;-TREC ratio quantification method for human thymic output measurement. \\u003cem\\u003eJ Immunol Methods\\u003c/em\\u003e \\u003cstrong\\u003e352\\u003c/strong\\u003e, 111\\u0026ndash;117 (2010).\\u003c/li\\u003e\\n\\u003cli\\u003eBulnes-Ramos, \\u0026Aacute;. \\u003cem\\u003eet al.\\u003c/em\\u003e Factors associated with the humoral response after three doses of COVID-19 vaccination in kidney transplant recipients. \\u003cem\\u003eFront Immunol\\u003c/em\\u003e \\u003cstrong\\u003e14\\u003c/strong\\u003e, 1099079 (2023).\\u003c/li\\u003e\\n\\u003cli\\u003eO\\u0026rsquo;Callaghan, N. J. \\u0026amp; Fenech, M. A quantitative PCR method for measuring absolute telomere length. \\u003cem\\u003eBiol Proced Online\\u003c/em\\u003e \\u003cstrong\\u003e13\\u003c/strong\\u003e, 3 (2011).\\u003c/li\\u003e\\n\\u003cli\\u003eTang, D. \\u003cem\\u003eet al.\\u003c/em\\u003e SRplot: A free online platform for data visualization and graphing. \\u003cem\\u003ePLoS One\\u003c/em\\u003e \\u003cstrong\\u003e18\\u003c/strong\\u003e, e0294236 (2023).\\u003c/li\\u003e\\n\\u003cli\\u003eZhang, Y. \\u0026amp; Zhang, Z. The history and advances in cancer immunotherapy: understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. \\u003cem\\u003eCell Mol Immunol\\u003c/em\\u003e \\u003cstrong\\u003e17\\u003c/strong\\u003e, 807\\u0026ndash;821 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eTan, S., Day, D., Nicholls, S. J. \\u0026amp; Segelov, E. Immune Checkpoint Inhibitor Therapy in Oncology: Current Uses and Future Directions: JACC: CardioOncology State-of-the-Art Review. \\u003cem\\u003eJACC CardioOncol\\u003c/em\\u003e \\u003cstrong\\u003e4\\u003c/strong\\u003e, 579\\u0026ndash;597 (2022).\\u003c/li\\u003e\\n\\u003cli\\u003e\\u0026Ouml;meroğlu, G. \\u0026amp; Mşek, Ş. Immunotherapy in the Treatment of Cancer: Today and Tomorrow. \\u003cem\\u003eCurrent Molecular Biology Reports 2024 10:3\\u003c/em\\u003e \\u003cstrong\\u003e10\\u003c/strong\\u003e, 54\\u0026ndash;64 (2024).\\u003c/li\\u003e\\n\\u003cli\\u003ePrasad, V., Haslam, A. \\u0026amp; Olivier, T. Updated estimates of eligibility and response: Immune checkpoint inhibitors. \\u003cem\\u003eJournal of Clinical Oncology\\u003c/em\\u003e \\u003cstrong\\u003e42\\u003c/strong\\u003e, e14613\\u0026ndash;e14613 (2024).\\u003c/li\\u003e\\n\\u003cli\\u003eHolder, A. M. \\u003cem\\u003eet al.\\u003c/em\\u003e Defining clinically useful biomarkers of immune checkpoint inhibitors in solid tumours. \\u003cem\\u003eNature Reviews Cancer 2024 24:7\\u003c/em\\u003e \\u003cstrong\\u003e24\\u003c/strong\\u003e, 498\\u0026ndash;512 (2024).\\u003c/li\\u003e\\n\\u003cli\\u003eKrieg, C. \\u003cem\\u003eet al.\\u003c/em\\u003e High-dimensional single-cell analysis predicts response to anti-PD-1 immunotherapy. \\u003cem\\u003eNature Medicine 2018 24:2\\u003c/em\\u003e \\u003cstrong\\u003e24\\u003c/strong\\u003e, 144\\u0026ndash;153 (2018).\\u003c/li\\u003e\\n\\u003cli\\u003eSubrahmanyam, P. B. \\u003cem\\u003eet al.\\u003c/em\\u003e Distinct predictive biomarker candidates for response to anti-CTLA-4 and anti-PD-1 immunotherapy in melanoma patients. \\u003cem\\u003eJ Immunother Cancer\\u003c/em\\u003e \\u003cstrong\\u003e6\\u003c/strong\\u003e, 18 (2018).\\u003c/li\\u003e\\n\\u003cli\\u003eWistuba-Hamprecht, K. \\u003cem\\u003eet al.\\u003c/em\\u003e Peripheral CD8 effector memory type 1 T-cells correlate with outcome in ipilimumab-treated stage IV melanoma patients. \\u003cem\\u003eEur J Cancer\\u003c/em\\u003e \\u003cstrong\\u003e73\\u003c/strong\\u003e, 61 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003eMankor, J. M. \\u003cem\\u003eet al.\\u003c/em\\u003e Efficacy of nivolumab and ipilimumab in patients with malignant pleural mesothelioma is related to a subtype of effector memory cytotoxic T cells: Translational evidence from two clinical trials. \\u003cem\\u003eEBioMedicine\\u003c/em\\u003e \\u003cstrong\\u003e62\\u003c/strong\\u003e, 103040 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eKluger, H. M. \\u003cem\\u003eet al.\\u003c/em\\u003e Characterization of PD-L1 expression and associated T-cell infiltrates in metastatic melanoma samples from variable anatomic sites. \\u003cem\\u003eClinical Cancer Research\\u003c/em\\u003e \\u003cstrong\\u003e21\\u003c/strong\\u003e, 3052\\u0026ndash;3060 (2015).\\u003c/li\\u003e\\n\\u003cli\\u003eDaud, A. I. \\u003cem\\u003eet al.\\u003c/em\\u003e Tumor immune profiling predicts response to anti\\u0026ndash;PD-1 therapy in human melanoma. \\u003cem\\u003eJ Clin Invest\\u003c/em\\u003e \\u003cstrong\\u003e126\\u003c/strong\\u003e, 3447\\u0026ndash;3452 (2016).\\u003c/li\\u003e\\n\\u003cli\\u003eWang, L. \\u003cem\\u003eet al.\\u003c/em\\u003e Hot and cold tumors: Immunological features and the therapeutic strategies. \\u003cem\\u003eMedComm (Beijing)\\u003c/em\\u003e \\u003cstrong\\u003e4\\u003c/strong\\u003e, e343 (2023).\\u003c/li\\u003e\\n\\u003cli\\u003eJeon, S. H. \\u003cem\\u003eet al.\\u003c/em\\u003e Dynamic changes in peripheral blood monocytes early after anti-PD-1 therapy predict clinical outcomes in hepatocellular carcinoma. \\u003cem\\u003eCancer Immunol Immunother\\u003c/em\\u003e \\u003cstrong\\u003e72\\u003c/strong\\u003e, 371 (2022).\\u003c/li\\u003e\\n\\u003cli\\u003eZhou, J. G. \\u003cem\\u003eet al.\\u003c/em\\u003e Prospective development and validation of a liquid immune profile-based signature (LIPS) to predict response of patients with recurrent/metastatic cancer to immune checkpoint inhibitors. \\u003cem\\u003eJ Immunother Cancer\\u003c/em\\u003e \\u003cstrong\\u003e9\\u003c/strong\\u003e, e001845 (2021).\\u003c/li\\u003e\\n\\u003cli\\u003eMartens, A. \\u003cem\\u003eet al.\\u003c/em\\u003e Baseline peripheral blood biomarkers associated with clinical outcome of advanced melanoma patients treated with ipilimumab. \\u003cem\\u003eClin Cancer Res\\u003c/em\\u003e \\u003cstrong\\u003e22\\u003c/strong\\u003e, 2908 (2016).\\u003c/li\\u003e\\n\\u003cli\\u003eKoch, S. \\u003cem\\u003eet al.\\u003c/em\\u003e Multiparameter flow cytometric analysis of CD4 and CD8 T cell subsets in young and old people. \\u003cem\\u003eImmunity and Ageing\\u003c/em\\u003e \\u003cstrong\\u003e5\\u003c/strong\\u003e, (2008).\\u003c/li\\u003e\\n\\u003cli\\u003eYu, X., Zhang, L., Chaudhry, A., Rapaport, A. S. \\u0026amp; Ouyang, W. Unravelling the heterogeneity and dynamic relationships of tumor-infiltrating T cells by single-cell RNA sequencing analysis. \\u003cem\\u003eJ Leukoc Biol\\u003c/em\\u003e \\u003cstrong\\u003e107\\u003c/strong\\u003e, 917\\u0026ndash;932 (2020).\\u003c/li\\u003e\\n\\u003cli\\u003eZheng, L. \\u003cem\\u003eet al.\\u003c/em\\u003e Pan-cancer single-cell landscape of tumor-infiltrating T cells. \\u003cem\\u003eScience (1979)\\u003c/em\\u003e \\u003cstrong\\u003e374\\u003c/strong\\u003e, (2021).\\u003c/li\\u003e\\n\\u003cli\\u003eSpeiser, D. E. \\u003cem\\u003eet al.\\u003c/em\\u003e Human CD8+ T cells expressing HLA-DR and CD28 show telomerase activity and are distinct from cytolytic effector T cells. \\u003cem\\u003eEur J Immunol\\u003c/em\\u003e \\u003cstrong\\u003e31\\u003c/strong\\u003e, 459\\u0026ndash;466 (2001).\\u003c/li\\u003e\\n\\u003cli\\u003eMiggelbrink, A. M. \\u003cem\\u003eet al.\\u003c/em\\u003e CD4 T-Cell exhaustion: Does it exist and what are its roles in cancer? \\u003cem\\u003eClinical Cancer Research\\u003c/em\\u003e \\u003cstrong\\u003e27\\u003c/strong\\u003e, 5742\\u0026ndash;5752 (2021).\\u003c/li\\u003e\\n\\u003cli\\u003eLingel, H. \\u0026amp; Brunner-Weinzierl, M. C. CTLA-4 (CD152): A versatile receptor for immune-based therapy. \\u003cem\\u003eSemin Immunol\\u003c/em\\u003e \\u003cstrong\\u003e42\\u003c/strong\\u003e, 101298 (2019).\\u003c/li\\u003e\\n\\u003cli\\u003eBrenchley, J. M. \\u003cem\\u003eet al.\\u003c/em\\u003e Expression of CD57 defines replicative senescence and antigen-induced apoptotic death of CD8+ T cells. \\u003cem\\u003eBlood\\u003c/em\\u003e \\u003cstrong\\u003e101\\u003c/strong\\u003e, 2711\\u0026ndash;2720 (2003).\\u003c/li\\u003e\\n\\u003cli\\u003eVerma, K. \\u003cem\\u003eet al.\\u003c/em\\u003e Human CD8+ CD57- TEMRA cells: Too young to be called \\u0026lsquo;old\\u0026rsquo;. \\u003cem\\u003ePLoS One\\u003c/em\\u003e \\u003cstrong\\u003e12\\u003c/strong\\u003e, e0177405 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003eLee, S. W. \\u003cem\\u003eet al.\\u003c/em\\u003e CD8+ TILs in NSCLC differentiate into TEMRA via a bifurcated trajectory: deciphering immunogenicity of tumor antigens. \\u003cem\\u003eJ Immunother Cancer\\u003c/em\\u003e \\u003cstrong\\u003e9\\u003c/strong\\u003e, e002709 (2021).\\u003c/li\\u003e\\n\\u003cli\\u003eBernal-Est\\u0026eacute;vez, D., S\\u0026aacute;nchez, R., Tejada, R. E. \\u0026amp; Parra-L\\u0026oacute;pez, C. Chemotherapy and radiation therapy elicits tumor specific T cell responses in a breast cancer patient. \\u003cem\\u003eBMC Cancer\\u003c/em\\u003e \\u003cstrong\\u003e16\\u003c/strong\\u003e, (2016).\\u003c/li\\u003e\\n\\u003cli\\u003eKuznetsova, M. \\u003cem\\u003eet al.\\u003c/em\\u003e Cytotoxic activity and memory T cell subset distribution of in vitro-stimulated CD8+ T cells specific for HER2/neu epitopes. \\u003cem\\u003eFront Immunol\\u003c/em\\u003e \\u003cstrong\\u003e10\\u003c/strong\\u003e, (2019).\\u003c/li\\u003e\\n\\u003cli\\u003eMelchiotti, R., Gracio, F., Kordasti, S., Todd, A. K. \\u0026amp; de Rinaldis, E. Cluster stability in the analysis of mass cytometry data. \\u003cem\\u003eCytometry Part A\\u003c/em\\u003e \\u003cstrong\\u003e91\\u003c/strong\\u003e, 73\\u0026ndash;84 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003eGarcia, S. \\u003cem\\u003eet al.\\u003c/em\\u003e Thoracic Oncology Clinical Trial Eligibility Criteria and Requirements Continue to Increase in Number and Complexity. \\u003cem\\u003eJournal of Thoracic Oncology\\u003c/em\\u003e \\u003cstrong\\u003e12\\u003c/strong\\u003e, 1489\\u0026ndash;1495 (2017).\\u003c/li\\u003e\\n\\u003cli\\u003eCooper, R. A., Chai, Y. \\u0026amp; Nieva, J. Effect of sponsor on enrollment criteria in non-small cell lung cancer clinical trials. \\u003cem\\u003eJ Cancer Policy\\u003c/em\\u003e \\u003cstrong\\u003e33\\u003c/strong\\u003e, 100336 (2022).\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":true,\"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\":\"info@researchsquare.com\",\"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\":\"Non-small cell lung cancer (NSCLC), immunotherapy, immune checkpoint inhibitors, pembrolizumab, single-cell mass cytometry, EMRA T cells, peripheral blood biomarkers, machine learning.\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-8656976/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-8656976/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"Background\\nMetastatic non-small cell lung cancer (NSCLC) exhibits high heterogeneity in response to immune checkpoint inhibitor (ICI) therapy. The identification of predictive biomarkers that are easily applicable and minimally invasive is therefore of great interest.\\nMethods\\nBaseline peripheral blood samples from real-world stage IV NSCLC patients treated with first-line pembrolizumab were analyzed using high-dimensional single-cell mass cytometry, clustering, and machine learning according to their response to ICI therapy. Functional immune characterization was additionally performed for validation.\\nResults\\nWe observed significantly higher frequencies of circulating CD4⁺ and CD8⁺ EMRA T cells, defined as terminally differentiated effector memory cells re-expressing CD45RA (CD45RA⁺CCR7⁻), in non-responder patients to programmed cell death protein 1 (PD-1) inhibitors. Upon in vitro stimulation, EMRA T cell subsets from non-responders showed reduced expression of activation markers and effector molecules, including CD25, CD69, IFN-γ, and GZMB. Based on these findings, we developed the CD4⁺ EMRA ImmunoPredict Score (CEIPS), a simplified predictive model that integrates activation and regulatory markers of CD4⁺ EMRA T cells. CEIPS stratified patients into Low- and High-Risk groups, with the latter showing significantly poorer progression-free and overall survival (p = 0.005 and p \\u003c 0.001, respectively).\\nConclusions\\nPeripheral CD4⁺ and CD8⁺ EMRA T cells are associated with anti-PD-1 immunotherapy response in metastatic NSCLC patients and represent clinically feasible blood-based biomarkers to improve patient stratification. Building on these findings, we developed the CEIPS score, which integrates these biomarkers and demonstrates predictive value for immunotherapy outcomes in NSCLC.\",\"manuscriptTitle\":\"Single-Cell Immune Profiling and Machine Learning Reveal a Predictive Immune Signature for Immunotherapy Response in NSCLC\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2026-01-27 06:48:26\",\"doi\":\"10.21203/rs.3.rs-8656976/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"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\":\"6c445413-686d-4d6f-b753-47fc121aefdf\",\"owner\":[],\"postedDate\":\"January 27th, 2026\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"posted\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2026-01-27T06:48:26+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2026-01-27 06:48:26\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-8656976\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-8656976\",\"identity\":\"rs-8656976\",\"version\":[\"v1\"]},\"buildId\":\"XKTyCvWXoU3ODBz1xrDgd\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}