Upfront whole blood transcriptional patterns in patients receiving immune checkpoint inhibitors associate with clinical outcome

Upfront whole blood transcriptional patterns in patients receiving immune checkpoint inhibitors associate with clinical outcome | 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 Upfront whole blood transcriptional patterns in patients receiving immune checkpoint inhibitors associate with clinical outcome Sara Hone Lopez, Stefan Loipfinger, Arkajyoti Bhattacharya, Mathilde Jalving, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6843569/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Sep, 2025 Read the published version in Cancer Immunology, Immunotherapy → Version 1 posted 10 You are reading this latest preprint version Abstract Whole blood (WB) transcriptomics offers a minimal-invasive method to assess patients’ immune system. This study aimed to identify transcriptional patterns in WB associated with clinical outcomes in patients treated with immune checkpoint inhibitors (ICIs). We performed RNA-sequencing on pre-treatment WB samples from 145 patients with advanced cancer. Additionally, we compiled a separate dataset of 14,085 WB transcriptomes from diverse health backgrounds from public repositories and applied consensus-independent component analysis (c-ICA) to identify transcriptional components (TCs). The biological processes represented by these TCs were elucidated using gene set enrichment analysis. The activity of the TCs was then quantified in the 145 WB profiles and analyzed for associations with tumor response, progression-free survival, and overall survival using univariate and multivariate analyses in a permutation framework. RNA-sequencing variant calling was performed, and the activity of the TCs was assessed in specific cell lineages using a single-cell immune cell atlas of the human hematopoietic system. c-ICA on 14,085 WB transcriptomes identified 1,262 distinct TCs representing various cellular processes. Of these, 18 TCs were associated with ≥1 outcome parameter, with three specifically linked to tumor response. Top genes in these three TCs included CCHCR1, TCF19, LTA, DDX39B, and PPPR1R18. RNA-sequencing variant calling and single-cell transcriptome projections revealed associations between these four TCs and germline variants. These findings support the potential of the identified WB-based transcriptional patterns to complement tumor characteristics in predictive and prognostic models for improved patient stratification. Immunotherapy Immune Checkpoint Inhibitors Biomarker Gene Expression Profiling Transcriptomics Figures Figure 1 Figure 2 INTRODUCTION Immune checkpoint inhibitors (ICIs) have improved patient outcomes across various tumor types. However, not all patients benefit from these treatments. [1] Currently, ICI treatment decisions for a select number of cancers incorporate three tumor-centric characteristics: i ) expression of the immune checkpoint programmed death-ligand 1 (PD-L1), ii ) tumor mutational burden, or iii ) the expression of DNA mismatch repair proteins. [2] These characteristics are assessed in tumor specimens, necessitating invasive procedures. Increasing evidence suggests that characteristics of a patient’s immune system also influence tumor response to ICIs. [3,4] Consequently, there is growing interest in minimally invasive methods enabling profiling patients’ immune characteristics. Whole blood (WB) transcriptomics is one such methodology, capable of identifying transcriptional patterns that reflect the cellular states of blood cells. A few studies with small patient cohorts or very limited gene panels have identified WB-derived transcriptional patterns associated with tumor response, progression-free survival (PFS), or overall survival (OS) in patients treated with ICIs. [5-8] These studies analyzed gene expression by examining the bulk RNA extracted from all cells in a WB sample. Consequently, measured gene expression levels reflect the aggregate transcriptional activity pattern across all sample cellular processes. This can obscure more subtle transcriptional patterns associated with antitumor activity. Consensus-independent component analysis (c-ICA) addresses this limitation by separating the cumulative transcriptional patterns into statistically independent transcriptional components (TCs). These TCs capture both prominent and subtle transcriptional patterns reflecting specific underlying cellular processes, as described previously. [9,10] Furthermore, the activity of these TCs can be determined in WB transcriptomes. In this study, we aimed to identify transcriptional patterns in WB from patients with advanced cancer treated with ICIs (PRIMERO-cohort) that are associated with treatment outcomes. We performed RNA-sequencing (RNA-seq) on prospectively collected pre-treatment whole blood (WB) samples from these patients and identified a broad range of TCs indicative of diverse cellular processes in an independent dataset of WB transcriptomes from public repositories. We then tested these TCs for associations with tumor response, PFS, and/or OS in the PRIMERO-cohort and with germline variants and specific blood cell types. MATERIALS AND METHODS All statistical analyses were conducted with R version 3.6.2. Further detailed information on the methods is provided in the Supplementary Methods. PRIMERO-cohort The PRIMERO-cohort comprised 145 patients who received treatment with ICIs for advanced melanoma, CSCC, MCC, RCC, or NSCLC between 2018 and 2022 at the Department of Medical Oncology and Pulmonology, University Medical Center Groningen. Patients were either enrolled in the prospective biobanks Oncological Life Study Immunotherapy cohort (OncoLifeS-Immunotherapy) or the clinical cohort study of patients with melanoma and NSCLC receiving checkpoint inhibitors (POINTING). Approval from the Medical Ethics Review Committee (METc) of the University Medical Center Groningen was obtained for the sample collection and subsequent research activities conducted as part of OncoLifeS (METc 2010/109, Dutch Trial Register: NL7839) and POINTING (METc 2018/350, NCT04193956). This study adhered to the Declaration of Helsinki. All patients provided informed written consent for the collection, anonymized data sharing, and collaborative use of their samples. For all patients, PAXgene™ Blood RNA tubes were collected before the start of treatment with ICI. Clinicopathological parameters gathered included sex, age at baseline, primary tumor, ICI formulation administered at baseline, cessation of ICI therapy for over 12 weeks, and subsequent cancer treatments (chemotherapy, targeted therapy, radiotherapy, or cancer surgery) received within a three-month window before, during, or following the primary treatment. Additionally, tumor response to ICI, PFS, and OS were recorded. Tumor response was assessed via (i)RECISTv1.1 and the best response recorded from start of ICI until disease progression or the last timepoint of follow-up was documented. Patients demonstrating a complete or partial response were considered responders, those exhibiting disease progression were non-responders. Patients with stable disease were excluded from the tumor response analysis to focus on those with observable changes in tumor burden. PFS was defined as the interval from the initiation of ICI therapy until disease progression according to RECISTv1.1, clinical progression if progression occurred prior to response evaluation took place, or death from any cause, whichever occurred first. If evaluation of PFS was not possible (switch from ICI to another treatment before the first response evaluation took place in the absence of clinical progression, or follow-up occurred elsewhere) this too was documented, and patients were excluded from PFS analyses. The same was applied to OS analyses if follow-up took place elsewhere. OS was defined at the interval from the initiation of ICI therapy to death from any cause. In cases where no progression nor death occurred, PFS and OS data were censored to the time at which all response evaluations and survival data were gathered (November 2023). Final clinical data lock was done in September 2024. RNA-seq of whole blood from the PRIMERO-cohort WB samples from the PRIMERO-cohort underwent whole-transcriptome RNA-seq at GenomeScan B.V. (Leiden, the Netherlands). Library preparation followed the ‘NEBNext Ultra II Directional RNA Library Prep Kit for Illumina’ (NEB #E7760S/L). mRNA was isolated from total RNA using oligo-dT beads, fragmented, and converted to cDNA, followed by adaptor ligation and PCR amplification. Quality and yield were assessed with a Fragment Analyzer. Sequencing was performed on a NovaSeq6000 (software v1.8) with a starting DNA concentration of 1.1 nM, per the manufacturer's protocol. Acquisition of whole blood transcriptomes from public repositories: PUBLIC dataset To increase the likelihood of uncovering transcriptional patterns in WB, public human WB transcriptomes were sourced from the ARCHS4 v2.2 database (Ensembl GRCh38.p13, release 107). [11] We curated a diverse dataset by selecting only ‘paxgene’-annotated samples, applying automated extraction and quality control filters to ensure sample relevance and data integrity. Transcriptomes were normalized for library size using size factors followed by a log 2 (x+1) transformation. [12] To reduce platform-related technical variability of the transcriptomes, principal component analysis (PCA) was applied to the sample correlation matrix and the first principal component, representing dominant non-biological variation, was removed. The resulting dataset is referred to as the PUBLIC dataset. Pre-processing of whole blood RNA-seq data from the PRIMERO-cohort RNA-seq reads from the PRIMERO-cohort were quality-checked using FastQC v0.11.9. The RNA-seq data processing was fully aligned with the procedures applied to the PUBLIC dataset. Raw reads were quantified employing the ARCHS4 pipeline, using the Python package archs4py v0.2.6. [11] Only genes present in the PUBLIC dataset were retained. Transcriptomes were normalized using size factors based on the geometric mean of the genes from the PUBLIC dataset, followed by log 2 (x+1) transformation. PCA was applied to correct for platform-specific effects, and the first principal component was removed. This ensured consistent processing between PRIMERO and PUBLIC WB transcriptomes. Consensus-independent component analysis and projection of PRIMERO-cohort transcriptomes onto transcriptional components from the PUBLIC dataset c-ICA is a statistical method to decompose mixed multivariate signals into their constituent source signals [9]. This study applies c-ICA to the PUBLIC bulk transcriptomes (each containing measurements for genes), considered as mixed multivariate signals, to isolate unique transcriptional components (i.e., TCs), each capturing a transcriptional pattern. These patterns are indicative of distinct underlying biological processes. [10] Each TC comprises gene weights, representing the direction and magnitude of the effect of the underlying biological process on a gene's expression level. The activity of the TCs identified in the PUBLIC cohort was then determined in the PRIMERO transcriptomes using a cross-dataset projection approach. [10] Determining the biological processes captured by the transcriptional components To elucidate the biological processes captured by the TCs, two methods were applied. First, gene set enrichment analysis (GSEA) was performed on all TCs using 12 gene set collections from the Molecular Signatures Database (MSigDB) v2023.1.Hs. [13] Gene sets comprising 10-500 genes were tested for enrichment among genes with high positive (≥3) or negative (≤-3) TC weights using 2x2 Fisher’s Exact Test. The most significant enrichment p-value per TC was Z-transformed for consistency across gene set sizes, Bonferroni-corrected, and considered biologically enriched if p < 0.05. Second, some TCs contain a limited number of highly weighted genes, termed small gene set-driven TCs, which often lack broad gene set enrichment but still carry biological meaning [14] These TCs were identified by ranking gene weights ( W i ) and calculating weight gaps ( G i =W i /W i+1 ). A TC was classified as small gene set-driven if any G i for a gene with W i >3 exceeded 1.5. [14] Associations between transcriptional components and clinical parameters Associations between TCs and age were evaluated using Spearman's rank correlation. Multinomial log-linear models were used to obtain association between TC activity and sex and tumor type. In addition, the Kruskal-Wallis test was applied to assess differences in TC’s activity per clinical parameter. Associations between transcriptional components and baseline hematological parameters Baseline hemoglobin, lymphocyte, and thrombocyte counts were recorded for all PRIMERO-cohort patients. White blood cell counts and erythrocytes were available for 42 patients. Associations between the activity of the outcome-associated TCs and hematological parameters were evaluated using Spearman's rank correlation. Only associations with a correlation p-value below 0.05 were considered significant. Associations between transcriptional components and outcomes Univariate and multivariate analyses were performed for the PRIMERO-cohort regarding tumor response to ICIs, PFS, and OS. TCs with a median absolute deviation (MAD) score ≥3, indicating sufficient inter-sample variability, were included in the survival analyses. [15] A logistic regression model was used to assess associations between TCs and tumor response. Cox proportional hazards models evaluated associations between TCs, OS, and PFS. Sample size precluded analyses per individual tumor type. However, this was addressed in the multivariate analyses which included covariates tumor type sex, age, ICI formulation given at baseline, and other cancer therapies received within a three-month window before, during, or following the primary treatment. The analysis was conducted in a multivariate permutation framework with 10,000 permutations to control the false discovery rate at 5% with an 80% confidence level. The robustness of associations between TCs and outcome was determined by selecting a random 80% subsamples of the data with 1,000 permutations. The sign and strength of associations (via coefficients and -log10 p-values) were computed for each TC. These results were compared to the original dataset's associations, with consistency and alignment used as measures of robustness. RNA-seq variant calling RNA-seq variant calling followed the GATK4 best-practice pipeline using nf-core/rnavar v2.0.0dev. Reads from the PRIMERO-cohort underwent quality control with FASTQC v0.12.1 and were aligned to Ensembl GRCh38.13 (release 107) using STAR v2.7.10a in two-pass mode (150 bp read length). [16] Duplicates were marked, spliced reads processed (SplitNCigarReads), and base quality recalibrated using GATK v4.5.0. [17] Variants were called with HaplotypeCaller and filtered based on FisherStrand >30, QualByDepth <2, Phred <30, synonymous status, total coverage <4, alt read count <2, variant allele frequency <30%, and occurrence in 3, were tested for association with TC activity using the Mann-Whitney U test. Bonferroni-corrected p-values <0.05 were considered significant. This analysis aimed to identify genetic characteristics for the haplotype versions of the highest-weighted genes in these TCs. HLA genotyping with arcasHLA HLA genotyping was performed on PRIMERO RNA-seq data using arcasHLA v0.5.0 with mapped reads from the RNA-seq variant calling analysis and the IMGT/HLA database v3.34.0. [18] For each sample, the alleles of the genes HLA-A, HLA-B, HLA-C, HLA-DQB1, HLA-DQA1, and HLA-DRB1 were determined. Association between presence of HLA alleles and activity of small gene set-driven outcome-associated TCs were tested using the Mann-Whitney U test with Bonferroni correction per TC, considering results with corrected p < 0.05 as significant. This analysis identified potential co-occurence of specific HLA alleles with the haplotype versions of the highest-weighted genes in these TCs. Determining the activity of transcriptional components associated with outcomes in single-cell RNA-seq data The activity of outcome-associated TCs in specific cell types was determined using single-cell RNA-seq data of umbilical cord blood and bone marrow samples from the single-cell immune cell atlas of the human hematopoietic system (see Data Availability Statement). The blood and bone marrow datasets were processed separately. Empty droplets were excluded, and half the cells were randomly sampled to reduce computational load. Provided cell type annotations were used. Genes not expressed in any cell were removed. Raw counts per cell were normalized by dividing by the total counts per cell, scaling by 10,000, and applying a log(x+1) transformation. The resulting processed transcriptomic profiles of each cell were then projected onto the TCs as described above. RESULTS PRIMERO-cohort characteristics To identify transcriptional patterns in WB associated with outcome in patients receiving ICIs, we prospectively collected whole blood before ICI administration from 145 patients between 2018 and 2022. These patients were treated for advanced disease of various cancers, including cutaneous squamous cell cancer (CSCC, n =10), melanoma ( n =83), Merkel cell cancer (MCC, n =7), non-small cell lung cancer (NSCLC, n =17), and renal cell cancer (RCC, n =28). This heterogeneous cohort enables us to identify systemic immune characteristics associated with ICI response, progression-free survival (PFS), and overall survival (OS), independent of tumor type. Detailed patient characteristics are provided in Table 1 and Supplementary Table 1. The median follow-up time was 20 months. According to the response criteria in solid tumors ((i)RECISTv1.1), the best objective responses observed were complete response in 19 patients, partial response in 46, stable disease in 15, and progressive disease in 65. The median PFS and OS were 4.5 (95% CI 0-61) and 20 (95% CI 1-61.4) months, respectively. Identification of 1,262 transcriptional components in 14,085 whole blood transcriptomes To maximize the identification of transcriptional patterns involved in diverse biological processes with WB, we applied c-ICA to a comprehensive collection of WB samples from public repositories covering a broad range of biological contexts (PUBLIC dataset). The PUBLIC dataset comprised 14,085 WB transcriptomes, encompassing 23,115 coding genes from healthy individuals and patients with various diseases (Table 2). We identified 1,262 distinct TCs by applying c-ICA to this dataset (Fig 1A-B, Supplementary Data 1). To elucidate the cellular processes represented by these TCs, we conducted gene set enrichment analysis (GSEA), revealing that 361 out of the 1,262 TCs were enriched for at least one gene set (Fig 1C). Detailed enrichment results are provided in Supplementary Data 2. Many of these TCs ( n =1,239) comprised a limited number of genes with high gene weights. These were classified as 'small gene set-driven TCs', of which 339 still demonstrated enrichment for at least one gene set. Our findings indicate that the identified TCs reflect transcriptional patterns associated with a broad range of specific cellular processes. The activities of 18 transcriptional components are associated with tumor response, progression-free survival, and/or overall survival Of the 1,262 TCs in the WB transcriptomes of the PRIMERO-cohort (Fig 1B), 70 TCs were identified with sufficient inter-sample activity variability (Supplementary Fig 1). These were included in subsequent analyses (Fig 1C-F). Of the 70 TCs, the activity of 18 TCs was associated with tumor response, PFS, and/or OS, either through univariate analysis or multivariate analysis incorporating covariates sex, age, primary tumor type, ICI regimen, and other cancer therapies (Fig 2, Supplementary Fig 2). While sample size precluded individual tumor type analyses, multivariate analysis identified TCs associated with outcomes independently of tumor type. Further insights, including results from biological characterization with GSEA, top genes in small gene set-driven TCs, and associations with variants, are provided in Table 3, Supplementary Fig 3, Supplementary Data 3 and 4. Notably, higher activity of three TCs (TC73, TC100, TC1125) was associated with longer OS and for TC1125 was also with responders (i.e., complete or partial tumor response). Conversely, higher activity of 15 TCs demonstrated associations with worse clinical outcomes. Specifically, 13 were associated with shorter OS, of which four (TC19, TC792, TC793, TC849) also with shorter PFS. Higher activity of TC865 was associated with non-responders and shorter PFS. Notably, higher activity of TC1238 was associated with non-responders and with shorter PFS and OS (Fig 2A). The robustness of associations between TCs and outcome are provided in Supplementary Data 4 and Supplementary Figure 4. Additionally, the activity of TC1093, TC1125, TC1175, and TC1238 was linked to RNA-seq-inferred HLA genotypes, including HLA-A*01:01 , HLA-B*08:01, HLA-C*07:01 , and HLA-DRB1*03:01 . TC671 was linked to HLA-B*07:02 and TC865 was linked to HLA-B*08:01 and HLA-C*07:01 (Supplementary Fig 5, Supplementary Data 4). In summary, higher activity of three TCs reflecting immune activation pathways was associated with favorable clinical outcomes. In contrast, 10 TCs reflecting pathways related to immune regulation akin to immune exhaustion were associated with worse outcome. The activity of the 18 outcome-associated transcriptional components in the single-cell immune atlas of the human hematopoietic system Our next objective was to identify cell types with higher activity for any of the 18 TCs associated with tumor response, PFS, and/or OS by assessing their activity across specific cell lineages in the single-cell immune cell atlas of the human hematopoietic system. For example, TC73 activity was highest in memory and naïve B cells, while TC100 exhibited higher activity in natural killer cells type 1 and cytotoxic type 1 and 2 T cells (Supplementary Fig 6, Supplementary Data 4). TC73 and TC100 activity was positively correlated with baseline hemoglobin and lymphocyte levels in the PRIMERO-cohort (Supplementary Data 3). Additionally, TC1151 activity was higher in CD14+ type 1 monocytes compared to other cell lineages. Higher activity of three TCs (TC19, TC792, and TC793) was observed in erythroid type 1 and 2 cells, with activities inversely correlated with baseline hemoglobin levels in the PRIMERO-cohort. DISCUSSION In this study, we identified a broad range of TCs indicative of diverse cellular processes in WB transcriptomes from public repositories. The activity of several of these TCs in prospectively collected pre-treatment WB samples from patients treated with ICIs was associated with tumor response, PFS, and/or OS. In our study, we leveraged the largest cohort to date for generating whole-transcriptomic profiles from whole-blood samples obtained from patients treated with ICIs. This extensive dataset allows for a more comprehensive exploration of transcriptional patterns compared to previous studies with smaller cohorts or limited gene panels. [5-8] While higher activity of three TCs was linked to longer OS, 13 were associated with shorter OS. It is important to note that multiple factors beyond ICI, such as patient characteristics and additional treatments, influence OS. However, tumor response is predominantly driven by ICIs, making it a more direct measure of their efficacy. Therefore, a deeper understanding of the biological processes represented by the TCs associated with tumor response may lead to improved patient stratification and enhanced tumor response to immune checkpoint therapies. The following discussion focuses on deciphering the biological mechanisms captured by the three TCs (TC1238, TC1125, and TC865) specifically linked to tumor response. Higher TC1238 activity was associated with a lack of tumor response, shorter PFS, and shorter OS in univariate and multivariate analyses. TC1238 included two haplotype alleles of CCHCR1 and TCF19 , genes that share a bidirectional promoter and carry the highest weights within this TC. [19]CRISPR knock-out screens of CCHCR1 in primary human CD8+ T cells and TCF19 in primary human T cells have shown increased T cell proliferation. [20, 21] CD8+ T cell proliferation in the blood has been associated with better tumor response to ICIs. [22, 23] The functional effects of the top single nucleotide polymorphism (SNP) variants associated with TC1238 activity remain unclear. However, CCHCR1 SNPs (rs1576, rs130079) have been associated with susceptibility to psoriasis, [24, 25] and SNPs rs1576 and TCF19 rs7750641 with type 1 diabetes mellitus. [26] In vitro studies have shown that CCHCR1 promotes steroidogenesis in cultured steroidogenic cells. [27, 28] In vivo data in mice suggest tumors may suppress anticancer immunity by promoting intratumoral steroidogenesis by T cells. [29] Therefore, CCHCR1 may indirectly contribute to tumor immune evasion. In summary, higher activity of TC1238 could be linked to biological processes that raise the threshold for CD8+ T cell proliferation after stimulus, which may explain its association with worse outcomes. Higher activity of TC865, with haplotype alleles of LTA and DDX39B carrying the highest weights, was associated with a lack of tumor response and shorter PFS. LTA is a cytokine from the tumor necrosis factor (TNF) family, and in vitro studies in Reed-Sternberg cells have shown that LTA knockout severely reduced PD-L1 and PD-L2 expression. [30] Similar to TCF19 , CRISPR knockout screens of LTA and DDX39B in primary T cells demonstrated increased T cell proliferation. [21] Additionally, CRISPR activation screens in primary T cells suggest that LTA inhibits cytokine production. [31] LTA binds to TNF receptor 1 (TNFR1) and 2 (TNFR2). [32] Interestingly, whereas TNFR1 signaling promotes inflammation and cell death, TNFR2 supports the opposite effects. [32, 33] Thus, LTA binding to TNFR1 and TNFR2 may promote both T cell death and immunosuppression. The specific mechanisms governing these interactions in the context of ICI remain unclear, and the functional impact of the top SNPs associated with TC865 activity is yet to be determined. In summary, higher TC865 activity may be related to mechanisms that hinder tumor response to ICIs by reducing T cell proliferation and effector function. In contrast to TC1238 and TC865, higher activity of TC1125, which included haplotype alleles for PPPR1R18 with the highest weights, was associated with better tumor response and longer OS. CRISPR screens in mice and Jurkat T cells showed that PPPR1R18 knockout reduces T cell's effector functions [34] Additionally, we found that higher TC1125 activity was associated with the absence of the PPPR1R18 SNP rs9262143, which has been linked to several autoimmune diseases. [35, 36] Therefore, TC1125 might be associated with biological processes that enhance T cell effector functions. We observed that TC865, TC1125, and TC1238 contained haplotype alleles of genes mapped to the MHC super-locus in the highly polymorphic genomic region 6p21. This genomic region encodes genes central to various immune processes, including antigen presentation, T cell differentiation, and the regulation of both innate and adaptive immune responses. [37, 38] SNPs in this region have also been linked to susceptibility to autoimmune diseases. [37-39] Specifically, lower TC1125 activity, and higher TC865 and TC1238 activity was associated with the HLA A1-B8-DR3-DQ2 haplotype, also known as the ancestral haplotype (AH8.1). This haplotype is associated with MHC class I and is linked to a predominance of type 2 cytokine response over type 1. [39] Type 1 cytokines promote cellular immune responses against cancer, [40] while type 2 cytokines support humoral immunity, reduce inflammation driven by type 1 cytokines, and assist tissue repair. [41, 42] An adequate balance between type 1 and type 2 cytokines is essential for an adequate anticancer immune response. [41, 43] Therefore, the immunoregulatory pathways associated with lower TC1125 activity and higher activity of TC865, and TC1238 likely represent mechanisms of immune evasion. For many outcome-associated TCs, higher activity was observed in specific cell types within the single-cell atlas of the human hematopoietic system. For instance, TC792 and TC793 exhibited higher activity in type 1 and 2 erythroid cells. However, for the four TCs discussed above, no distinct cell type showed significantly higher activity. This suggests that germline SNP variants may be driving the variation in TC activity that we detected in the 145 patients. The 1,262 distinct TCs identified from the 14,085 WB transcriptomes offer a valuable resource for researchers worldwide. By determining the activity of these TCs in their own WB samples, researchers can explore associations with phenotypes of interest. The integration of large-scale public datasets enables the detection of both pronounced and subtle transcriptional patterns, even in relatively small cohorts, helping to uncover meaningful associations that may otherwise remain hidden in isolated studies. Validation of our results in a larger prospective cohort is crucial to fully assess the predictive and prognostic value of the 18 identified TCs. Nevertheless, our findings support the potential of the identified WB-based transcriptional patterns to complement tumor characteristics in predictive and prognostic models for improved patient stratification. Declarations ACKNOWLEDGMENTS We would like to thank all the patients who selflessly took the time to donate samples for this project. FUNDING STATEMENT This work was supported by a grant from the UMCG Kanker Researchfonds. AUTHOR CONTRIBUTIONS J.J.H. and R.S.N.F. conceived and designed the study. S.H.L., S.L., and R.S.N.F. curated the datasets, performed the data analyses, and wrote the manuscript. S.H.L. and S.L. developed the figures used in this study. A.B., M.J., S.F.O., T.J.N.H., M.B., E.G.E.V., J.J.H., and R.S.N.F. provided critical input and contributed to the writing and editing of the manuscript. All authors have read and approve the final version of the manuscript. DATA AVAILABILITY STATEMENT Public WB transcriptomes for the PUBLIC dataset were acquired from the ARCHS4 database (https://maayanlab.cloud/archs4/). PUBLIC and PRIMERO-cohort RNA-seq read count data, TC activities, and all codes required to reproduce the results are available at https://zenodo.org/records/14902818. RNA-seq variant calling pipeline is available at: https://github.com/nf-core/rnavar/. PRIMERO-cohort raw whole-transcriptome RNA-seq data is available upon reasonable request. Single-cell RNA-seq data of umbilical cord blood and bone marrow samples from the single-cell immune cell atlas of the human hematopoietic system are available at: https://explore.data.humancellatlas.org/projects/cc95ff89-2e68-4a08-a234-480eca21ce79. The composition of the TC and GSEA results are available in Supplementary Data 1 and 2, respectively. The source data for the figures can be found in Supplementary Data 4. DECLARATION OF INTEREST STATEMENT S.H.L., S.L., A.B., R.S.N.F., and J.J.H. declare no competing interests. M.J. reports Institutional financial support for advisory boards from Pierre Fabre, Astra Zeneca, and Regeneron; all unrelated to the submitted work. J.T.N.H. reports financial support for advisory boards or DSMB from Roche, Bristol Myers Squibb, MSD, and Pfizer; institutional financial support for clinical trials or contracted research grants from Roche, Bristol Myers Squibb, and Astra Zeneca; other financial or non-financial interests as principal investigator of clinical studies for Astra Zeneca, GSK, Novartis, Merck Serono, Roche, Bristol Myers Squibb, and Amgen; all unrelated to the submitted work. S.F.O. reports Institutional financial support for advisory boards/consultancy from Bristol Myers Squibb, Genmab, Merck, Travel Congress Management B.V.; institutional financial support for clinical trials or contracted research grants from Celldex, Pfizer, Novartis, and Merck; received non-financial support from Celldex, Pfizer, Novartis; is the principal investigator of EORTC 2120 for which EORTC has received funding from Merck; and is a member of steering and safety monitoring committee of ALX Oncology (unpaid); all unrelated to the submitted work. E.G.E.V. reports Institutional financial support for advisory boards/consultancy from NSABP, Daiichi Sankyo, and Crescendo Biologics, and institutional financial support for clinical trials or contracted research grants from Amgen, Genentech, Roche, Servier, Regeneron, Bayer, and Crescendo Biologics; all unrelated to the submitted work. M.B. received grants from the Dutch Cancer Society (KWF), the European Research Council (ERC), Health Holland (HH), Mendus, BioNovion, Aduro Biotech, Vicinivax, Genmab, and IMMIOS (all paid to the institute); received non-financial support from BioNTech, Surflay Nanotec, and Merck Sharp & Dohme; is stock option holder in Sairopa; all unrelated to the submitted work. REFERENCES Haslam A, Kim MS, Prasad V (2021) Updated estimates of eligibility for and response to genome-targeted oncology drugs among US cancer patients, 2006-2020. Ann Oncol 32:926-32. doi: 10.1016/j.annonc.2021.04.003 Yamaguchi H, Hsu JM, Sun L, et al (2024) Advances and prospects of biomarkers for immune checkpoint inhibitors. Cell Rep Med 5:101621. doi: 10.1016/j.xcrm.2024.101621 Wang X, Muzaffar J, Kirtane K, et al (2022) T cell repertoire in peripheral blood as a potential biomarker for predicting response to concurrent cetuximab and nivolumab in head and neck squamous cell carcinoma. J Immunother Cancer 10: e004512. doi: 10.1136/jitc-2022-004512 Naranbhai V, Viard M, Dean M, et al (2022) HLA-A*03 and response to immune checkpoint blockade in cancer: an epidemiological biomarker study. Lancet Oncol 23:172-84. doi: 10.1016/S1470-2045(21)00582-9 Aiba T, Hattori C, Sugisaka J, et al (2021) Gene expression signatures as candidate biomarkers of response to PD-1 blockade in non-small cell lung cancers. PLoS One 16:e0260500. doi: 10.1371/journal.pone.0260500 van Wilpe S, Wosika V, Ciarloni L, et al (2021) Whole blood transcriptome profiling identifies DNA replication and cell cycle regulation as early marker of response to anti-PD-1 in patients with urothelial cancer. Cancers (Basel) 13:4660. doi: 10.3390/cancers13184660 Nagumo Y, Kandori S, Kojima T, et al (2022) Whole blood gene expression profiles correlate with response to immune checkpoint inhibitors in patients with metastatic renal cell carcinoma. Cancers (Basel) 14:6207. doi: 10.3390/cancers14246207 Friedlander P, Wassmann K, Christenfeld AM, et al (2017) Whole blood RNA transcript-based models can predict clinical response in two large independent clinical studies of patients with advanced melanoma treated with the checkpoint inhibitor, tremelimumab. J Immunother Cancer 2017;5:67. doi: 10.1186/s40425-017-0272-z Bhattacharya A, Bense RD, Urzúa-Traslaviña CG, et al (2020) Transcriptional effects of copy number alterations in a large set of human cancers. Nat Commun 11:715. doi: 10.1038/s41467-020-14605-5 Knapen DG, Hone Lopez S, de Groot DJA, et al (2024) Independent transcriptional patterns reveal biological processes associated with disease-free survival in early colorectal cancer. Commun Med (Lond). 4:79. doi: 10.1038/s43856-024-00504-z Lachmann A, Torre D, Keenan AB, et al (2018) Massive mining of publicly available RNA-seq data from human and mouse. Nat Commun 9:1366. doi: 10.1038/s41467-018-03751-6 Love MI, Huber W, Anders S (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. doi: 10.1186/s13059-014-0550-8 Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P (2015) The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1:417-25. doi: 10.1016/j.cels.2015.12.004 Kairov U, Cantini L, Greco A, et al (2017) Determining the optimal number of independent components for reproducible transcriptomic data analysis. BMC Genomics 18:712. doi: 10.1186/s12864-017-4112-9 Chung N, Zhang XD, Kreamer A, et al (2008) Median absolute deviation to improve hit selection for genome-scale RNAi screens. J Biomol Screen 13:149-58. doi: 10.1177/1087057107312035 Dobin A, Davis CA, Schlesinger F, et al (2013) STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29:15-21. doi: 10.1093/bioinformatics/bts635 Van der Auwera GA, Carneiro MO, Hartl C, et al (2013) From FastQ data to high‐confidence variant calls: the genome analysis toolkit best practices pipeline. Curr Protoc Bioinformatics 43:1.10.1-11.10.33. doi: 10.1002/0471250953.bi1110s43 Orenbuch R, Filip I, Comito D, Shaman J, Pe'er I, Rabadan R (2020) arcasHLA: high-resolution HLA typing from RNAseq. Bioinformatics 36:33-40. doi: 10.1093/bioinformatics/btz474 Ling YH, Chen Y, Leung KN, Chan KM, Liu WK (2023) Cell cycle regulation of the psoriasis associated gene CCHCR1 by transcription factor E2F1 PloS One 18:e0294661. doi: 10.1371/journal.pone.0294661 Shifrut E, Carnevale J, Tobin V, et al (2018) Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell 2018;175:1958-1971.e1 doi: 10.1016/j.cell.2018.10.024 Carnevale J, Shifrut E, Kale N, et al (2022) RASA2 ablation in T cells boosts antigen sensitivity and long-term function. Nature 609:174-82. doi: 10.1038/s41586-022-05126-w Kim KH, Cho J, Ku BM, et al (2019) The first-week proliferative response of peripheral blood PD-1+CD8+ T cells predicts the response to anti-PD-1 therapy in solid tumors. Clin Cancer Res 25:2144-54. doi: 10.1158/1078-0432.CCR-18-1449 Kamphorst AO, Pillai RN, Yang S, et al (2017) Proliferation of PD-1+ CD8 T cells in peripheral blood after PD-1-targeted therapy in lung cancer patients. Proc Natl Acad Sci U S A 114:4993-98. doi: 10.1073/pnas.1705327114 Zhou F, Shen C, Xu J, et al (2016) Epigenome-wide association data implicates DNA methylation-mediated genetic risk in psoriasis. Clin Epigenetics 8:131. doi: 10.1186/s13148-016-0297 Prieto-Pérez R, Solano-López G, Cabaleiro T, et al (2015) Polymorphisms associated with age at onset in patients with moderate-to-severe plaque psoriasis. J Immunol Res 2015:101879. doi: 10.1155/2015/101879 Cheung YH, Watkinson J, Anastassiou D (2011) Conditional meta-analysis stratifying on detailed HLA genotypes identifies a novel type 1 diabetes locus around TCF19 in the MHC. Hum Genet 129:161-76. doi: 10.1007/s00439-010-0908-2 Sugawara T, Shimizu H, Hoshi N, Nakajima A, Fujimoto S (2003) Steroidogenic acute regulatory protein-binding protein cloned by a yeast two-hybrid system. J Biol Chem 278:42487–94. doi: 10.1074/jbc.M302291200 Tiala I, Suomela S, Huuhtanen J, et al (2020) The CCHCR1 (HCR) gene is relevant for skin steroidogenesis and downregulated in cultured psoriatic keratinocytes. J Mol Med 85:589–601. doi: 10.1007/s00109-006-0155-0 Mahata B, Pramanik J, van der Weyden L, et al (2020) Tumors induce de novo steroid biosynthesis in T cells to evade immunity. Nat Comm 11:3588. doi: 10.1038/s41467-020-17339-6 von Hoff L, Kärgel E, Franke V, et al (2019) Autocrine LTA signaling drives NF-κB and JAK-STAT activity and myeloid gene expression in Hodgkin lymphoma. Blood 133:1489-94. doi: 10.1182/blood-2018-08-871293 Schmidt R, Steinhart Z, Layeghi M, et al (2022) CRISPR activation and interference screens decode stimulation responses in primary human T cells. Science 375:eabj4008. doi: 10.1126/science.abj4008 Cheng Y, Liu Y, Xu D, et al (2024) An engineered TNFR1-selective human lymphotoxin-alpha mutant delivered by an oncolytic adenovirus for tumor immunotherapy. Biochim Biophys Acta Mol Basis Dis 1870:167122. doi: 10.1016/j.bbadis.2024.167122 Beldi G, Khosravi M, Abdelgawad ME, et al (2020) TNFα/TNFR2 signaling pathway: an active immune checkpoint for mesenchymal stem cell immunoregulatory function. Stem Cell Res Ther 11:281. doi: 10.1186/s13287-020-01740-5 Lee HS, Sun H, Lagarrigue F, et al (2022) Phostensin enables lymphocyte integrin activation and population of peripheral lymphoid organs. J Exp Med 219:e20211637. doi: 10.1084/jem.20211637 Armstrong DL, Zidovetzki R, Alarcón-Riquelme ME, et al (2014) GWAS identifies novel SLE susceptibility genes and explains the association of the HLA region. Genes Immun 15:347-54. doi: 10.1038/gene.2014.23 Hakonarson H, Grant SF, Bradfield JP, et al (2007) A genome-wide association study identifies KIAA0350 as a type 1 diabetes gene. Nature 448:591-94. doi: 10.1038/nature06010 Matzaraki V, Kumar V, Wijmenga C, Zhernakova A (2017) The MHC locus and genetic susceptibility to autoimmune and infectious diseases. Genome Biol 18:76. doi: 10.1186/s13059-017-1207-1 Kulski JK, Suzuki S, Shiina T (2022) Human leukocyte antigen super-locus: nexus of genomic supergenes, SNPs, indels, transcripts, and haplotypes. Hum Genome Var 9:49. doi: 10.1038/s41439-022-00226-5 Gambino CM, Aiello A, Accardi G, Caruso C, Candore G (2018) Autoimmune diseases and 8.1 ancestral haplotype: An update. HLA 92:137-43. doi: 10.1111/tan.13305 Rozeman EA, Hoefsmit EP, Reijers ILM, et al (2021) Survival and biomarker analyses from the OpACIN-neo and OpACIN neoadjuvant immunotherapy trials in stage III melanoma. Nat Med 27:256-63. doi: 10.1038/s41591-020-01211-7 Jou E (2023) Type 1 and type 2 cytokine-mediated immune orchestration in the tumour microenvironment and their therapeutic potential. Explor Target Antitumor Ther 4:474-97. doi: 10.37349/etat.2023.00146 Wynn T (2015) Type 2 cytokines: mechanisms and therapeutic strategies. Nat Rev Immunol 15:271–82. doi: 10.1038/nri3831 Gao X, Wang X, Yang Q, et al (2015) Tumoral expression of IL-33 inhibits tumor growth and modifies the tumor microenvironment through CD8+ T and NK cells. J Immunol 194:438-45. doi: 10.4049/jimmunol.1401344 Tables Tables 1 to 3 are available in the Supplementary Files section Additional Declarations Competing interest reported. S.H.L., S.L., A.B., R.S.N.F., and J.J.H. declare no competing interests. M.J. reports Institutional financial support for advisory boards from Pierre Fabre, Astra Zeneca, and Regeneron; all unrelated to the submitted work. J.T.N.H. reports financial support for advisory boards or DSMB from Roche, Bristol Myers Squibb, MSD, and Pfizer; institutional financial support for clinical trials or contracted research grants from Roche, Bristol Myers Squibb, and Astra Zeneca; other financial or non-financial interests as principal investigator of clinical studies for Astra Zeneca, GSK, Novartis, Merck Serono, Roche, Bristol Myers Squibb, and Amgen; all unrelated to the submitted work. S.F.O. reports Institutional financial support for advisory boards/consultancy from Bristol Myers Squibb, Genmab, Merck, Travel Congress Management B.V.; institutional financial support for clinical trials or contracted research grants from Celldex, Pfizer, Novartis, and Merck; received non-financial support from Celldex, Pfizer, Novartis; is the principal investigator of EORTC 2120 for which EORTC has received funding from Merck; and is a member of steering and safety monitoring committee of ALX Oncology (unpaid); all unrelated to the submitted work. E.G.E.V. reports Institutional financial support for advisory boards/consultancy from NSABP, Daiichi Sankyo, and Crescendo Biologics, and institutional financial support for clinical trials or contracted research grants from Amgen, Genentech, Roche, Servier, Regeneron, Bayer, and Crescendo Biologics; all unrelated to the submitted work. M.B. received grants from the Dutch Cancer Society (KWF), the European Research Council (ERC), Health Holland (HH), Mendus, BioNovion, Aduro Biotech, Vicinivax, Genmab, and IMMIOS (all paid to the institute); received non-financial support from BioNTech, Surflay Nanotec, and Merck Sharp & Dohme; is stock option holder in Sairopa; all unrelated to the submitted work. Supplementary Files SupplementaryData1.xlsx SupplementaryData2.xlsx SupplementaryData3.xlsx SupplementaryData4.xlsx Supplementaryfileswithlabelsrevised.pdf Tables.docx Cite Share Download PDF Status: Published Journal Publication published 11 Sep, 2025 Read the published version in Cancer Immunology, Immunotherapy → Version 1 posted Editorial decision: Revision requested 25 Jun, 2025 Reviews received at journal 23 Jun, 2025 Reviews received at journal 22 Jun, 2025 Reviewers agreed at journal 12 Jun, 2025 Reviewers agreed at journal 10 Jun, 2025 Reviewers agreed at journal 10 Jun, 2025 Reviewers invited by journal 10 Jun, 2025 Editor assigned by journal 09 Jun, 2025 Submission checks completed at journal 09 Jun, 2025 First submitted to journal 07 Jun, 2025 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-6843569","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":469515285,"identity":"19ec7337-b044-4817-95cc-17886c49de72","order_by":0,"name":"Sara Hone Lopez","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Sara","middleName":"Hone","lastName":"Lopez","suffix":""},{"id":469515287,"identity":"9c09134d-5f8b-47c7-a86e-009a3c813f10","order_by":1,"name":"Stefan Loipfinger","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Stefan","middleName":"","lastName":"Loipfinger","suffix":""},{"id":469515288,"identity":"0b3f3d50-ae60-487b-8df5-7423b883e0ad","order_by":2,"name":"Arkajyoti Bhattacharya","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Arkajyoti","middleName":"","lastName":"Bhattacharya","suffix":""},{"id":469515289,"identity":"da60cc76-f919-4939-b1b5-82aedde3cb6d","order_by":3,"name":"Mathilde Jalving","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Mathilde","middleName":"","lastName":"Jalving","suffix":""},{"id":469515290,"identity":"5c2ced34-4b33-489b-8295-257dd8105e1c","order_by":4,"name":"Sjoukje F. Oosting","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Sjoukje","middleName":"F.","lastName":"Oosting","suffix":""},{"id":469515291,"identity":"3dfd8feb-7edb-458f-bcaa-42313ef35e11","order_by":5,"name":"Jeroen T. N. Hiltermann","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Jeroen","middleName":"T. N.","lastName":"Hiltermann","suffix":""},{"id":469515292,"identity":"5ded2ce4-6ab0-4c5d-a82d-191e7ab2aae5","order_by":6,"name":"Marco de Bruyn","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Marco","middleName":"","lastName":"de Bruyn","suffix":""},{"id":469515293,"identity":"8f73663a-111d-401c-bade-4a5e352f41a8","order_by":7,"name":"Elisabeth G.E. de Vries","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Elisabeth","middleName":"G.E.","lastName":"de Vries","suffix":""},{"id":469515294,"identity":"c5e68adf-863c-4864-a4bb-6db32485718e","order_by":8,"name":"Jacco Juri de Haan","email":"","orcid":"","institution":"University of Groningen","correspondingAuthor":false,"prefix":"","firstName":"Jacco","middleName":"Juri","lastName":"de Haan","suffix":""},{"id":469515295,"identity":"557584b2-3abf-4499-a5dc-221c3f7d4cc7","order_by":9,"name":"Rudolf S.N. Fehrmann","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4klEQVRIie3RsQrCMBCA4YMMXS67UK2vcNJBRPFZLIKTQt3cdGo3Z30IQSgUx0CHDhZ1FFwUH8FRB0/URTTqJpJ/Klc+LiEAJtPv5gBdP2q3gf+WuHfSug3oc5K8J+WBtRU+kFO2596hO1t6kzBdCV9D8gpJjIDcyrAT2eNs48VZ2xcjDckBgkA4edNMTm0ZMFljQ6CW8MEQqM8kOspg8QkBupAGZTLmLYqJpfSE/yZIVOItcVUGTZfvAomWWOFujz0q8pZoI4N6IU7Ty+Q1AQGgHl4BSWnA06ztt8JkMpn+uzO27kYS6wEw+AAAAABJRU5ErkJggg==","orcid":"","institution":"University of Groningen","correspondingAuthor":true,"prefix":"","firstName":"Rudolf","middleName":"S.N.","lastName":"Fehrmann","suffix":""}],"badges":[],"createdAt":"2025-06-07 15:08:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6843569/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6843569/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00262-025-04155-4","type":"published","date":"2025-09-11T15:57:01+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":84528712,"identity":"8c75fd7b-9e0b-4dc7-b7bd-c38e433823c0","added_by":"auto","created_at":"2025-06-13 05:45:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1600096,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIntegrated approach to identifying blood-related transcriptional components associated with outcomes in patients receiving immune checkpoint inhibitors\u003c/strong\u003e. We prospectively collected pre-immune checkpoint inhibitor patient whole blood (WB) samples and clinical data (PRIMERO-cohort). \u003cstrong\u003eA \u003c/strong\u003eAdditionally, we built a dataset with \u0026gt;14,000 WB transcriptomes of public repositories of healthy individuals and patients with various conditions (PUBLIC). We applied consensus-independent component analysis (c-ICA) to the PUBLIC dataset to obtain statistically independent transcriptional components (TCs), capturing prominent and subtle blood-related transcriptional patterns. \u003cstrong\u003eB \u003c/strong\u003eWe performed RNA-seq on the PRIMERO-cohort samples and determined the activity of PUBLIC TCs in the PRIMERO-cohort. \u003cstrong\u003eC \u003c/strong\u003eThe biological processes captured by these TCs were characterized through gene set enrichment analyses.\u003cstrong\u003e D \u003c/strong\u003eNext, we identified TCs associated with clinical parameters, tumor response, PFS, and OS in the PRIMERO-cohort. \u003cstrong\u003eE \u003c/strong\u003eIn addition, HLA genotyping was performed on the PRIMERO-cohort RNA-seq data. \u003cstrong\u003eF\u003c/strong\u003e Finally, we investigated in which cell types the TCs associated with outcome had higher activity in the single-cell immune cell atlas of the human hematopoietic system. ICI, immune checkpoint inhibitor; NR, non-responder; OS, overall survival; PFS, progression-free survival; R, responder; TC, transcriptional component.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/88f22baac9205ddc48a6ebca.png"},{"id":84528714,"identity":"d549b992-594c-4d91-a4b3-172977a28efc","added_by":"auto","created_at":"2025-06-13 05:45:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2856880,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBiology captured by the 18 transcriptional components associated with overall survival, progression-free survival, and tumor response to immune checkpoint inhibitors.\u003c/strong\u003e \u003cstrong\u003eA\u003c/strong\u003e The heatmaps in the left section illustrate the associations between transcriptional components (TCs) and clinical outcomes. TCs for which higher activity was associated with better clinical outcomes are indicated in green, while those associated with worse clinical outcomes are noted in red. The shade of green or red reflects the significance of the association. The TCs for which the null hypothesis was rejected in the permutation test are marked with a white asterisk. The central heatmap displays the results of GSEA focusing on Hallmark gene sets. The heatmap on the right section of the figure shows the association between TC activity and clinical parameters. \u003cstrong\u003eB \u003c/strong\u003eGenes with an absolute weight greater than three for the small gene set-driven TCs are listed below the heatmap, and the specific haplotype alleles for each gene are shown above. Genes with a positive weight greater than three are indicated in red, and genes with a negative weight less than three are indicated in blue. The intensity of the blue or green color represents the magnitude of the gene weight in the TC. ICI, immune checkpoint inhibitor; OS, overall survival; PFS, progression-free survival; TC, transcriptional component.\u003cstrong\u003eBiology captured by the 18 transcriptional components associated with overall survival, progression-free survival, and tumor response to immune checkpoint inhibitors.\u003c/strong\u003e \u003cstrong\u003eA\u003c/strong\u003e The heatmaps in the left section illustrate the associations between transcriptional components (TCs) and clinical outcomes. TCs for which higher activity was associated with better clinical outcomes are indicated in green, while those associated with worse clinical outcomes are noted in red. The shade of green or red reflects the significance of the association. The TCs for which the null hypothesis was rejected in the permutation test are marked with a white asterisk. The central heatmap displays the results of GSEA focusing on Hallmark gene sets. The heatmap on the right section of the figure shows the association between TC activity and clinical parameters. \u003cstrong\u003eB \u003c/strong\u003eGenes with an absolute weight greater than three for the small gene set-driven TCs are listed below the heatmap, and the specific haplotype alleles for each gene are shown above. Genes with a positive weight greater than three are indicated in red, and genes with a negative weight less than three are indicated in blue. The intensity of the blue or green color represents the magnitude of the gene weight in the TC. ICI, immune checkpoint inhibitor; OS, overall survival; PFS, progression-free survival; TC, transcriptional component.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/45d9791d875dc0613af6349c.png"},{"id":91358973,"identity":"5efcbed0-6f56-4cbd-ac84-4a52d562b084","added_by":"auto","created_at":"2025-09-15 16:02:04","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5148497,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/0d980e26-544c-4fbd-b447-fee1d23aec3e.pdf"},{"id":84528723,"identity":"b668364e-9848-4be1-a832-854704962f05","added_by":"auto","created_at":"2025-06-13 05:45:29","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":427373407,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryData1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/8bc452096d53ae8657a61979.xlsx"},{"id":84528722,"identity":"d466e1f2-e677-48ca-902d-e7e6ecf70902","added_by":"auto","created_at":"2025-06-13 05:45:18","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":192442111,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryData2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/5c9596cb106c1d48339d8c3b.xlsx"},{"id":84528713,"identity":"8d6d2581-1bd6-4189-8c07-fb2d15a40ca0","added_by":"auto","created_at":"2025-06-13 05:45:08","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":31169,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryData3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/337ee66855316816cd5679e9.xlsx"},{"id":84528716,"identity":"e757f051-49e0-4166-a83a-7e41ff9ffd3e","added_by":"auto","created_at":"2025-06-13 05:45:13","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":88790458,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryData4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/ef2e588ca8e08a8e997f6ecc.xlsx"},{"id":84528719,"identity":"e4b339cf-b3ef-455b-87bc-16ac53ab0c2e","added_by":"auto","created_at":"2025-06-13 05:45:14","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":103534189,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfileswithlabelsrevised.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/4e035129581bb1a3cb441f48.pdf"},{"id":84528711,"identity":"30dc25bd-e41a-416a-b4f8-56f8356b180c","added_by":"auto","created_at":"2025-06-13 05:45:08","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":26158,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-6843569/v1/617798f3a86ebb8f65e8fbc2.docx"}],"financialInterests":"Competing interest reported. S.H.L., S.L., A.B., R.S.N.F., and J.J.H. declare no competing interests. M.J. reports Institutional financial support for advisory boards from Pierre Fabre, Astra Zeneca, and Regeneron; all unrelated to the submitted work. J.T.N.H. reports financial support for advisory boards or DSMB from Roche, Bristol Myers Squibb, MSD, and Pfizer; institutional financial support for clinical trials or contracted research grants from Roche, Bristol Myers Squibb, and Astra Zeneca; other financial or non-financial interests as principal investigator of clinical studies for Astra Zeneca, GSK, Novartis, Merck Serono, Roche, Bristol Myers Squibb, and Amgen; all unrelated to the submitted work. S.F.O. reports Institutional financial support for advisory boards/consultancy from Bristol Myers Squibb, Genmab, Merck, Travel Congress Management B.V.; institutional financial support for clinical trials or contracted research grants from Celldex, Pfizer, Novartis, and Merck; received non-financial support from Celldex, Pfizer, Novartis; is the principal investigator of EORTC 2120 for which EORTC has received funding from Merck; and is a member of steering and safety monitoring committee of ALX Oncology (unpaid); all unrelated to the submitted work. E.G.E.V. reports Institutional financial support for advisory boards/consultancy from NSABP, Daiichi Sankyo, and Crescendo Biologics, and institutional financial support for clinical trials or contracted research grants from Amgen, Genentech, Roche, Servier, Regeneron, Bayer, and Crescendo Biologics; all unrelated to the submitted work. M.B. received grants from the Dutch Cancer Society (KWF), the European Research Council (ERC), Health Holland (HH), Mendus, BioNovion, Aduro Biotech, Vicinivax, Genmab, and IMMIOS (all paid to the institute); received non-financial support from BioNTech, Surflay Nanotec, and Merck Sharp \u0026 Dohme; is stock option holder in Sairopa; all unrelated to the submitted work.","formattedTitle":"Upfront whole blood transcriptional patterns in patients receiving immune checkpoint inhibitors associate with clinical outcome","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eImmune checkpoint inhibitors (ICIs) have improved patient outcomes across various tumor types. However, not all patients benefit from these treatments. [1] Currently, ICI treatment decisions for a select number of cancers incorporate three tumor-centric characteristics: \u003cem\u003ei\u003c/em\u003e) expression of the immune checkpoint programmed death-ligand 1 (PD-L1), \u003cem\u003eii\u003c/em\u003e) tumor mutational burden, or \u003cem\u003eiii\u003c/em\u003e) the expression of DNA mismatch repair proteins. [2] These characteristics are assessed in tumor specimens, necessitating invasive procedures.\u003c/p\u003e\n\u003cp\u003eIncreasing evidence suggests that characteristics of a patient’s immune system also influence tumor response to ICIs. [3,4] Consequently, there is growing interest in minimally invasive methods enabling profiling patients’ immune characteristics. Whole blood (WB) transcriptomics is one such methodology, capable of identifying transcriptional patterns that reflect the cellular states of blood cells.\u003c/p\u003e\n\u003cp\u003eA few studies with small patient cohorts or very limited gene panels have identified WB-derived transcriptional patterns associated with tumor response, progression-free survival (PFS), or overall survival (OS) in patients treated with ICIs. [5-8] These studies analyzed gene expression by examining the bulk RNA extracted from all cells in a WB sample. Consequently, measured gene expression levels reflect the aggregate transcriptional activity pattern across all sample cellular processes. This can obscure more subtle transcriptional patterns associated with antitumor activity.\u003c/p\u003e\n\u003cp\u003eConsensus-independent component analysis (c-ICA) addresses this limitation by separating the cumulative transcriptional patterns into statistically independent transcriptional components (TCs). These TCs capture both prominent and subtle transcriptional patterns reflecting specific underlying cellular processes, as described previously. [9,10] Furthermore, the activity of these TCs can be determined in WB transcriptomes.\u003c/p\u003e\n\u003cp\u003eIn this study, we aimed to identify transcriptional patterns in WB from patients with advanced cancer treated with ICIs (PRIMERO-cohort) that are associated with treatment outcomes. We performed RNA-sequencing (RNA-seq) on prospectively collected pre-treatment whole blood (WB) samples from these patients and identified a broad range of TCs indicative of diverse cellular processes in an independent dataset of WB transcriptomes from public repositories. We then tested these TCs for associations with tumor response, PFS, and/or OS in the PRIMERO-cohort and with germline variants and specific blood cell types.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cp\u003eAll statistical analyses were conducted with R version 3.6.2. Further detailed information on the methods is provided in the Supplementary Methods.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePRIMERO-cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe PRIMERO-cohort comprised 145 patients who received treatment with ICIs\u0026nbsp;for advanced melanoma, CSCC, MCC, RCC, or NSCLC between 2018 and 2022 at the Department of Medical Oncology and Pulmonology, University Medical Center Groningen. Patients were either enrolled in the prospective biobanks Oncological Life Study Immunotherapy cohort (OncoLifeS-Immunotherapy) or the clinical cohort study of patients with melanoma and NSCLC receiving checkpoint inhibitors (POINTING). Approval from the Medical Ethics Review Committee (METc) of the University Medical Center Groningen was obtained for the sample collection and subsequent research activities conducted as part of OncoLifeS (METc 2010/109, Dutch Trial Register: NL7839) and POINTING (METc 2018/350, NCT04193956). This study adhered to the Declaration of Helsinki. All patients provided informed written consent for the collection, anonymized data sharing, and collaborative use of their samples.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;For all patients, PAXgene™ Blood RNA tubes were collected before the start of treatment\u0026nbsp;with ICI. Clinicopathological parameters gathered included sex, age at baseline, primary tumor, ICI formulation administered at baseline, cessation of ICI therapy for over 12 weeks, and subsequent cancer treatments (chemotherapy, targeted therapy, radiotherapy, or cancer surgery) received within a three-month window before, during, or following the primary treatment. Additionally, tumor response to ICI, PFS, and OS were recorded.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTumor response was assessed via (i)RECISTv1.1 and the best response recorded from start of ICI until disease progression or the last timepoint of follow-up\u003c/p\u003e\n\u003cp\u003ewas documented. Patients demonstrating a complete or partial response were considered responders, those exhibiting disease progression were non-responders. Patients with stable disease were excluded from the tumor response analysis to focus on those with observable changes in tumor burden. PFS was defined as the interval from the initiation of ICI therapy until disease progression according to RECISTv1.1, clinical progression if progression occurred prior to response evaluation took place, or death from any cause, whichever occurred first. If evaluation of PFS was not possible (switch from ICI to another treatment before the first response evaluation took place in the absence of clinical progression, or follow-up occurred elsewhere) this too was documented, and patients were excluded from PFS analyses. The same was applied to OS analyses if follow-up took place elsewhere. OS was defined at the interval from the initiation of ICI therapy to death from any cause. In cases where no progression nor death occurred, PFS and OS data were censored to the time at which all response evaluations and survival data were gathered (November 2023). Final clinical data lock was done in September 2024.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRNA-seq of whole blood from the PRIMERO-cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWB samples from the PRIMERO-cohort underwent whole-transcriptome RNA-seq at GenomeScan B.V. (Leiden, the Netherlands). Library preparation followed the ‘NEBNext Ultra II Directional RNA Library Prep Kit for Illumina’ (NEB #E7760S/L). mRNA was isolated from total RNA using oligo-dT beads, fragmented, and converted to cDNA, followed by adaptor ligation and PCR amplification. Quality and yield were assessed with a Fragment Analyzer. Sequencing was performed on a NovaSeq6000 (software v1.8) with a starting DNA concentration of 1.1 nM, per the manufacturer's protocol.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcquisition of whole blood transcriptomes from public repositories: PUBLIC dataset\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo increase the likelihood of uncovering transcriptional patterns in WB, public human WB transcriptomes were sourced from the ARCHS4 v2.2 database (Ensembl GRCh38.p13, release 107). [11] We curated a diverse dataset by selecting only ‘paxgene’-annotated samples, applying automated extraction and quality control filters to ensure sample relevance and data integrity.\u003c/p\u003e\n\u003cp\u003eTranscriptomes were normalized for library size using size factors followed by a log\u003csub\u003e2\u003c/sub\u003e(x+1) transformation. [12] To reduce platform-related technical variability of the transcriptomes, principal component analysis (PCA) was applied to the sample correlation matrix and the first principal component, representing dominant non-biological variation, was removed. The resulting dataset is referred to as the PUBLIC dataset.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePre-processing of whole blood RNA-seq data from the PRIMERO-cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRNA-seq reads from the PRIMERO-cohort were quality-checked using FastQC v0.11.9. The RNA-seq data processing was fully aligned with the procedures applied to the PUBLIC dataset. Raw reads were quantified employing the ARCHS4 pipeline, using the Python package archs4py v0.2.6. [11] Only genes present in the PUBLIC dataset were retained. Transcriptomes were normalized using size factors based on the geometric mean of the genes from the PUBLIC dataset, followed by log\u003csub\u003e2\u003c/sub\u003e(x+1) transformation. PCA was applied to correct for platform-specific effects, and the first principal component was removed. This ensured consistent processing between PRIMERO and PUBLIC WB transcriptomes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsensus-independent component analysis and projection of PRIMERO-cohort transcriptomes onto transcriptional components from the PUBLIC dataset\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ec-ICA is a statistical method to decompose mixed multivariate signals into their constituent source signals [9]. This study applies c-ICA to the PUBLIC bulk transcriptomes (each containing measurements for\u0026nbsp;\u0026nbsp;\u0026nbsp;genes), considered as mixed multivariate signals, to isolate unique transcriptional components (i.e., TCs), each capturing a transcriptional pattern. These patterns are indicative of distinct underlying biological processes. [10] Each TC comprises\u0026nbsp;\u0026nbsp;\u0026nbsp;gene weights, representing the direction and magnitude of the effect of the underlying biological process on a gene's expression level. The activity of the TCs identified in the PUBLIC cohort was then determined in the PRIMERO transcriptomes using a cross-dataset projection approach. [10]\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetermining the biological processes captured by the transcriptional components\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo elucidate the biological processes captured by the TCs, two methods were applied. First, gene set enrichment analysis (GSEA) was performed on all TCs using 12 gene set collections from the Molecular Signatures Database (MSigDB) v2023.1.Hs. [13] Gene sets comprising 10-500 genes were tested for enrichment among genes with high positive (≥3) or negative (≤-3) TC weights using 2x2 Fisher’s Exact Test. The most significant enrichment p-value per TC was Z-transformed for consistency across gene set sizes, Bonferroni-corrected, and considered biologically enriched if p \u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003eSecond, some TCs contain a limited number of highly weighted genes, termed small gene set-driven TCs, which often lack broad gene set enrichment but still carry biological meaning [14] These TCs were identified by ranking gene weights (\u003cem\u003eW\u003csub\u003ei\u003c/sub\u003e\u003c/em\u003e) and calculating\u0026nbsp;weight gaps (\u003cem\u003eG\u003csub\u003ei\u003c/sub\u003e=W\u003csub\u003ei\u003c/sub\u003e/W\u003csub\u003ei+1\u003c/sub\u003e\u003c/em\u003e). A TC was classified as small gene set-driven if any\u0026nbsp;\u003cem\u003eG\u003csub\u003ei\u003c/sub\u003e\u003c/em\u003e for a gene with\u0026nbsp;\u003cem\u003eW\u003csub\u003ei\u003c/sub\u003e\u0026gt;3\u003c/em\u003e exceeded 1.5. [14]\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAssociations between transcriptional components and clinical parameters\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAssociations between TCs and age were evaluated using Spearman's rank correlation. Multinomial log-linear models were used to obtain association between TC activity and sex and tumor type. In addition, the Kruskal-Wallis test was applied to assess differences in TC’s activity per clinical parameter.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAssociations between transcriptional components and baseline hematological parameters\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBaseline hemoglobin, lymphocyte, and thrombocyte counts were recorded for all PRIMERO-cohort patients. White blood cell counts and erythrocytes were available for 42 patients. Associations between the activity of the outcome-associated TCs and hematological parameters were evaluated using Spearman's rank correlation. Only associations with a correlation p-value below 0.05 were considered significant.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAssociations between transcriptional components and outcomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUnivariate and multivariate analyses were performed for the PRIMERO-cohort regarding tumor response to ICIs, PFS, and OS. TCs with a median absolute deviation (MAD) score ≥3, indicating sufficient inter-sample variability, were included in the survival analyses. [15] A logistic regression model was used to assess associations between TCs and tumor response. Cox proportional hazards models evaluated associations between TCs, OS, and PFS.\u0026nbsp;Sample size precluded analyses per individual tumor type. However, this was addressed in the multivariate analyses which included covariates tumor type\u0026nbsp;sex, age, ICI formulation given at baseline, and other cancer therapies received within a three-month window before, during, or following the primary treatment. The analysis was conducted in a multivariate permutation framework with 10,000 permutations to control the false discovery rate at 5% with an 80% confidence level.\u003c/p\u003e\n\u003cp\u003eThe robustness of associations between TCs and outcome was determined by selecting a random 80% subsamples of the data with 1,000 permutations. The sign and strength of associations (via coefficients and -log10 p-values) were computed for each TC. These results were compared to the original dataset's associations, with consistency and alignment used as measures of robustness.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRNA-seq variant calling\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRNA-seq variant calling followed the GATK4 best-practice pipeline using nf-core/rnavar v2.0.0dev. Reads from the PRIMERO-cohort underwent quality control with FASTQC v0.12.1 and were aligned to Ensembl GRCh38.13 (release 107) using STAR v2.7.10a in two-pass mode (150 bp read length). [16] Duplicates were marked, spliced reads processed (SplitNCigarReads), and base quality recalibrated using GATK v4.5.0. [17] Variants were called with HaplotypeCaller and filtered based on FisherStrand \u0026gt;30, QualByDepth \u0026lt;2, Phred \u0026lt;30, synonymous status, total coverage \u0026lt;4, alt read count \u0026lt;2, variant allele frequency \u0026lt;30%, and occurrence in \u0026lt;10 samples.\u003c/p\u003e\n\u003cp\u003eFor small gene set-driven outcome-associated TC, only variants in genes with absolute weights \u0026gt;3, were tested for association with TC activity using the Mann-Whitney U test. Bonferroni-corrected p-values \u0026lt;0.05 were considered significant. This analysis aimed to identify genetic characteristics for the haplotype versions of the highest-weighted genes in these TCs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHLA genotyping with arcasHLA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHLA genotyping was performed on PRIMERO RNA-seq data using arcasHLA v0.5.0 with mapped reads from the RNA-seq variant calling analysis and the IMGT/HLA database v3.34.0. [18] For each sample, the alleles of the genes HLA-A, HLA-B, HLA-C, HLA-DQB1, HLA-DQA1, and HLA-DRB1 were determined. Association between presence of HLA alleles and activity of small gene set-driven outcome-associated TCs were tested using the Mann-Whitney U test with Bonferroni correction per TC, considering results with corrected p\u0026nbsp;\u0026lt;\u0026nbsp;0.05 as significant. This analysis identified potential co-occurence of specific HLA alleles with the haplotype versions of the highest-weighted genes in these TCs.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetermining the activity of transcriptional components associated with outcomes in single-cell RNA-seq data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe activity of outcome-associated TCs in specific cell types was determined using single-cell RNA-seq data of umbilical cord blood and bone marrow samples from the single-cell immune cell atlas of the human hematopoietic system (see Data Availability Statement).\u003c/p\u003e\n\u003cp\u003eThe blood and bone marrow datasets were processed separately. Empty droplets were excluded, and half the cells were randomly sampled to reduce computational load. Provided cell type annotations were used. Genes not expressed in any cell were removed. Raw counts per cell were normalized by dividing by the total counts per cell, scaling by 10,000, and applying a log(x+1) transformation. The resulting processed transcriptomic profiles of each cell were then projected onto the TCs as described above.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cstrong\u003ePRIMERO-cohort characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo\u0026nbsp;identify transcriptional patterns in WB associated with outcome in patients receiving ICIs, we prospectively collected whole blood before ICI administration from 145 patients between 2018 and 2022. These patients were treated for advanced disease of various cancers, including\u0026nbsp;cutaneous squamous cell cancer (CSCC, \u003cem\u003en\u003c/em\u003e=10), melanoma\u0026nbsp;(\u003cem\u003en\u003c/em\u003e=83), Merkel cell cancer (MCC,\u0026nbsp;\u003cem\u003en\u003c/em\u003e=7), non-small cell lung cancer (NSCLC,\u0026nbsp;\u003cem\u003en\u003c/em\u003e=17), and renal cell cancer (RCC,\u0026nbsp;\u003cem\u003en\u003c/em\u003e=28). This heterogeneous cohort enables us to identify systemic immune characteristics associated with ICI response, progression-free survival (PFS), and overall survival (OS), independent of tumor type.\u0026nbsp;Detailed patient characteristics are provided in Table 1 and Supplementary Table 1. The median follow-up time was 20 months. According to the response criteria in solid tumors ((i)RECISTv1.1), the best objective responses observed were complete response in 19 patients, partial response in 46, stable disease in 15, and progressive disease in 65. The median PFS and OS were 4.5 (95% CI 0-61) and 20 (95% CI 1-61.4) months, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIdentification of 1,262 transcriptional components in 14,085 whole blood transcriptomes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo maximize the identification of transcriptional patterns involved in diverse biological processes with WB, we applied c-ICA to a comprehensive collection of WB samples from public repositories covering a broad range of biological contexts (PUBLIC dataset).\u0026nbsp;The\u0026nbsp;PUBLIC dataset comprised 14,085 WB transcriptomes, encompassing 23,115 coding genes from healthy individuals and\u0026nbsp;patients with various diseases (Table 2). We identified 1,262 distinct TCs by applying c-ICA to this dataset (Fig 1A-B, Supplementary Data 1). To elucidate the cellular processes represented by these TCs, we conducted gene set enrichment analysis (GSEA), revealing that 361 out of the 1,262 TCs were enriched for at least one gene set (Fig 1C). Detailed enrichment results are provided in Supplementary Data 2. Many of these TCs (\u003cem\u003en\u003c/em\u003e=1,239) comprised a limited number of genes with high gene weights. These were classified as 'small gene set-driven TCs', of which 339 still demonstrated enrichment for at least one gene set.\u003c/p\u003e\n\u003cp\u003eOur findings indicate that the identified TCs reflect transcriptional patterns associated with a broad range of specific cellular processes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe activities of 18 transcriptional components are associated with tumor response, progression-free survival, and/or overall survival\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOf the 1,262 TCs in the WB transcriptomes of the PRIMERO-cohort (Fig 1B), 70 TCs were identified with sufficient inter-sample activity variability (Supplementary Fig 1). These were included in subsequent analyses (Fig 1C-F). Of the 70 TCs, the activity of 18 TCs was associated with tumor response, PFS, and/or OS, either through univariate analysis or multivariate analysis incorporating covariates sex, age, primary tumor type, ICI regimen, and other cancer therapies (Fig 2, Supplementary Fig 2). While sample size precluded individual tumor type analyses, multivariate analysis identified TCs associated with outcomes independently of tumor type. Further insights, including results from biological characterization with GSEA, top genes in small gene set-driven TCs, and associations with variants, are provided in Table 3, Supplementary Fig 3, Supplementary Data 3 and 4.\u003c/p\u003e\n\u003cp\u003eNotably, higher activity of three TCs (TC73, TC100, TC1125) was associated with longer OS and for TC1125 was also with responders (i.e., complete or partial tumor response). Conversely, higher activity of 15 TCs demonstrated associations with worse clinical outcomes. Specifically, 13 were associated with shorter OS, of which four (TC19, TC792, TC793, TC849) also with shorter PFS. Higher activity of TC865 was associated with non-responders and shorter PFS. Notably, higher activity of TC1238 was associated with non-responders and with shorter PFS and OS (Fig 2A).\u0026nbsp;The robustness of associations between TCs and outcome are provided in Supplementary Data 4 and Supplementary Figure 4.\u003c/p\u003e\n\u003cp\u003eAdditionally, the activity of TC1093, TC1125, TC1175, and TC1238 was linked to RNA-seq-inferred HLA genotypes, including \u003cem\u003eHLA-A*01:01\u003c/em\u003e, \u003cem\u003eHLA-B*08:01, HLA-C*07:01\u003c/em\u003e, and \u003cem\u003eHLA-DRB1*03:01\u003c/em\u003e. TC671 was linked to \u003cem\u003eHLA-B*07:02\u003c/em\u003e and TC865 was linked to \u003cem\u003eHLA-B*08:01\u003c/em\u003e and \u003cem\u003eHLA-C*07:01\u003c/em\u003e (Supplementary Fig 5, Supplementary Data 4).\u003c/p\u003e\n\u003cp\u003eIn summary, higher activity of three TCs reflecting immune activation pathways was associated with favorable clinical outcomes. In contrast, 10 TCs reflecting pathways related to immune regulation akin to immune exhaustion were associated with worse outcome.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe activity of the 18 outcome-associated transcriptional components in the single-cell immune atlas of the human hematopoietic system\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur next objective was to identify cell types with higher activity for any of the 18 TCs associated with\u0026nbsp;tumor response, PFS, and/or OS by assessing their activity across specific cell lineages in the\u0026nbsp;single-cell immune cell atlas of the human hematopoietic system.\u003c/p\u003e\n\u003cp\u003eFor example, TC73 activity was highest in memory and naïve B cells, while TC100 exhibited higher activity in natural killer cells type 1 and cytotoxic type 1 and 2 T cells (Supplementary Fig 6, Supplementary Data 4). TC73 and TC100 activity was positively correlated with baseline hemoglobin and lymphocyte levels in the PRIMERO-cohort (Supplementary Data 3). Additionally, TC1151 activity was higher in CD14+ type 1 monocytes compared to other cell lineages. Higher activity of three TCs (TC19, TC792, and TC793) was observed in erythroid type 1 and 2 cells, with activities inversely correlated with baseline hemoglobin levels in the PRIMERO-cohort.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we identified a broad range of TCs indicative of diverse cellular processes in WB transcriptomes from public repositories. The activity of several of these TCs in prospectively collected pre-treatment WB samples from patients treated with ICIs was associated with\u0026nbsp;tumor response, PFS, and/or OS.\u003c/p\u003e\n\u003cp\u003eIn our study, we leveraged the largest cohort to date for generating whole-transcriptomic profiles from whole-blood samples obtained from patients treated with ICIs. This extensive dataset allows for a more comprehensive exploration of transcriptional patterns compared to previous studies with smaller cohorts or limited gene panels. [5-8]\u003c/p\u003e\n\u003cp\u003eWhile higher activity of three TCs was linked to longer OS, 13 were associated with shorter OS. It is important to note that multiple factors beyond ICI, such as patient characteristics and additional treatments, influence OS. However, tumor response is predominantly driven by ICIs, making it a more direct measure of their efficacy. Therefore, a deeper understanding of the biological processes represented by the TCs associated with tumor response may lead to improved patient stratification and enhanced tumor response to immune checkpoint therapies. The following discussion focuses on deciphering the biological mechanisms captured by the three TCs (TC1238, TC1125, and TC865) specifically linked to tumor response.\u003c/p\u003e\n\u003cp\u003eHigher TC1238 activity was associated with a\u0026nbsp;lack of tumor response, shorter PFS, and shorter OS\u0026nbsp;in univariate and multivariate analyses. TC1238 included two haplotype alleles of \u003cem\u003eCCHCR1\u0026nbsp;\u003c/em\u003eand \u003cem\u003eTCF19\u003c/em\u003e, genes that share a bidirectional promoter and carry the highest weights within this TC. [19]CRISPR knock-out screens of \u003cem\u003eCCHCR1\u003c/em\u003e in primary human CD8+ T cells and \u003cem\u003eTCF19\u003c/em\u003e in primary human T cells have shown increased T cell proliferation. [20, 21] CD8+ T cell proliferation in the blood has been associated with better tumor response to ICIs. [22, 23] The functional effects of the top single nucleotide polymorphism (SNP) variants associated with TC1238 activity remain unclear. However, \u003cem\u003eCCHCR1\u003c/em\u003e SNPs (rs1576, rs130079) have been associated with susceptibility to psoriasis, [24, 25] and SNPs rs1576 and \u003cem\u003eTCF19\u003c/em\u003e rs7750641 with type 1 diabetes mellitus. [26] In vitro studies have shown that \u003cem\u003eCCHCR1\u003c/em\u003e promotes steroidogenesis in cultured steroidogenic cells. [27, 28] In vivo data in mice suggest tumors may suppress anticancer immunity by promoting intratumoral steroidogenesis by T cells. [29] Therefore, \u003cem\u003eCCHCR1\u003c/em\u003e may indirectly contribute to tumor immune evasion. In summary, higher activity of TC1238 could be linked to biological processes that raise the threshold for CD8+ T cell proliferation after stimulus, which may explain its association with worse outcomes.\u003c/p\u003e\n\u003cp\u003eHigher activity of TC865, with haplotype alleles of \u003cem\u003eLTA\u003c/em\u003e and\u0026nbsp;\u003cem\u003eDDX39B\u003c/em\u003e carrying the highest weights, was associated with a lack of tumor response and shorter PFS.\u0026nbsp;\u003cem\u003eLTA\u003c/em\u003e is a cytokine from the tumor necrosis factor (TNF) family, and in vitro studies in Reed-Sternberg cells have shown that \u003cem\u003eLTA\u003c/em\u003e knockout severely reduced PD-L1 and PD-L2 expression. [30] Similar to \u003cem\u003eTCF19\u003c/em\u003e,\u0026nbsp;CRISPR knockout screens of \u003cem\u003eLTA\u003c/em\u003e and \u003cem\u003eDDX39B\u003c/em\u003e in primary T cells demonstrated increased T cell proliferation. [21] Additionally, CRISPR activation screens in primary T cells suggest that \u003cem\u003eLTA\u003c/em\u003e inhibits cytokine production. [31] LTA binds to TNF receptor 1 (TNFR1) and 2 (TNFR2). [32] Interestingly, whereas TNFR1 signaling promotes inflammation and cell death, TNFR2 supports the opposite effects. [32, 33] Thus, \u003cem\u003eLTA\u003c/em\u003e binding to TNFR1 and TNFR2 may promote both T cell death and immunosuppression. The specific mechanisms governing these interactions in the context of\u0026nbsp;ICI\u0026nbsp;remain unclear, and\u0026nbsp;the functional impact of the top SNPs associated with TC865 activity is yet to be determined. In summary, higher TC865 activity may be related to mechanisms that hinder tumor response to ICIs by reducing T cell proliferation and effector function.\u003c/p\u003e\n\u003cp\u003eIn contrast to\u0026nbsp;TC1238 and TC865, higher activity of TC1125, which included haplotype alleles for \u003cem\u003ePPPR1R18\u003c/em\u003e with the highest weights, was associated with better tumor response and longer OS. CRISPR screens in mice and Jurkat T cells showed that \u003cem\u003ePPPR1R18\u003c/em\u003e knockout reduces T cell's effector functions [34] Additionally, we found that higher TC1125 activity was associated with the absence of the PPPR1R18 SNP rs9262143, which has been linked to several autoimmune diseases. [35, 36] Therefore, TC1125 might be associated with biological processes that enhance T cell effector functions.\u003c/p\u003e\n\u003cp\u003eWe observed that TC865, TC1125, and TC1238 contained haplotype alleles of genes mapped to the MHC super-locus in the highly polymorphic genomic region 6p21. This genomic region encodes genes central to various immune processes, including antigen presentation, T cell differentiation, and the regulation of both innate and adaptive immune responses. [37, 38] SNPs in this region have also been linked to susceptibility to autoimmune diseases. [37-39] Specifically, lower TC1125 activity, and higher TC865 and TC1238 activity was associated with the HLA A1-B8-DR3-DQ2 haplotype, also known as the ancestral haplotype (AH8.1). This haplotype is associated with MHC class I and is linked to a predominance of type 2 cytokine response over type 1. [39] Type 1 cytokines promote cellular immune responses against cancer, [40] while type 2 cytokines support humoral immunity, reduce inflammation driven by type 1 cytokines, and assist tissue repair. [41, 42] An adequate balance between type 1 and type 2 cytokines is essential for an adequate anticancer immune response. [41, 43] Therefore, the immunoregulatory pathways associated with lower TC1125 activity and higher activity of TC865, and TC1238 likely represent mechanisms of immune evasion.\u003c/p\u003e\n\u003cp\u003eFor many outcome-associated TCs, higher activity was observed in specific cell types within the single-cell atlas of the human hematopoietic system. For instance, TC792 and TC793 exhibited higher activity in type 1 and 2 erythroid cells. However, for the four TCs discussed above, no distinct cell type showed significantly higher activity. This suggests that germline SNP variants may be driving the variation in TC activity that we detected in the 145 patients.\u003c/p\u003e\n\u003cp\u003eThe 1,262 distinct TCs identified from the 14,085 WB transcriptomes offer a valuable resource for researchers worldwide. By determining the activity of these TCs in their own WB samples, researchers can explore associations with phenotypes of interest. The integration of large-scale public datasets enables the detection of both pronounced and subtle transcriptional patterns, even in relatively small cohorts, helping to uncover meaningful associations that may otherwise remain hidden in isolated studies.\u003c/p\u003e\n\u003cp\u003eValidation of our results in a larger prospective cohort is crucial to fully assess the predictive and prognostic value of the 18 identified TCs. Nevertheless, our\u0026nbsp;findings support the potential of the identified WB-based transcriptional patterns to complement tumor characteristics in predictive and prognostic models for improved patient stratification.\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eACKNOWLEDGMENTS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to thank all the patients who selflessly took the time to donate samples for this project.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFUNDING STATEMENT\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by a grant from the UMCG Kanker Researchfonds.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAUTHOR CONTRIBUTIONS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.J.H. and R.S.N.F. conceived and designed the study. S.H.L., S.L., and R.S.N.F. curated the datasets, performed the data analyses, and wrote the manuscript. S.H.L. and S.L. developed the figures used in this study. A.B., M.J., S.F.O., T.J.N.H., M.B., E.G.E.V., J.J.H., and R.S.N.F. provided critical input and contributed to the writing and editing of the manuscript. All authors have read and approve the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDATA AVAILABILITY STATEMENT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePublic WB transcriptomes for the PUBLIC dataset were acquired from the ARCHS4 database (https://maayanlab.cloud/archs4/). PUBLIC and PRIMERO-cohort RNA-seq read count data, TC activities, and all codes required to reproduce the results are available at https://zenodo.org/records/14902818. RNA-seq variant calling pipeline is available at: https://github.com/nf-core/rnavar/. PRIMERO-cohort raw whole-transcriptome RNA-seq data is available upon reasonable request. Single-cell RNA-seq data of umbilical cord blood and bone marrow samples from the single-cell immune cell atlas of the human hematopoietic system are available at: https://explore.data.humancellatlas.org/projects/cc95ff89-2e68-4a08-a234-480eca21ce79. The composition of the TC and GSEA results are available in Supplementary Data 1 and 2, respectively. The source data for the figures can be found in Supplementary Data 4.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDECLARATION OF INTEREST STATEMENT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eS.H.L., S.L., A.B., R.S.N.F., and J.J.H. declare no competing interests. M.J. reports Institutional financial support for advisory boards from Pierre Fabre, Astra Zeneca, and Regeneron; all unrelated to the submitted work. J.T.N.H. reports financial support for advisory boards or DSMB from Roche, Bristol Myers Squibb, MSD, and Pfizer; institutional financial support for clinical trials or contracted research grants from Roche, Bristol Myers Squibb, and Astra Zeneca; other financial or non-financial interests as principal investigator of clinical studies for Astra Zeneca, GSK, Novartis, Merck Serono, Roche, Bristol Myers Squibb, and Amgen; all unrelated to the submitted work. S.F.O. reports Institutional financial support for advisory boards/consultancy from Bristol Myers Squibb, Genmab, Merck, Travel Congress Management B.V.; institutional financial support for clinical trials or contracted research grants from Celldex, Pfizer, Novartis, and Merck; received non-financial support from Celldex, Pfizer, Novartis; is the\u0026nbsp;principal investigator of EORTC 2120 for which EORTC has received funding from Merck; and is a member of steering and safety monitoring committee of ALX Oncology (unpaid); all unrelated to the submitted work. E.G.E.V. reports Institutional financial support for advisory boards/consultancy from NSABP, Daiichi Sankyo, and Crescendo Biologics, and institutional financial support for clinical trials or contracted research grants from Amgen, Genentech, Roche, Servier, Regeneron, Bayer, and Crescendo Biologics; all unrelated to the submitted work. M.B. received grants from the Dutch Cancer Society (KWF), the European Research Council (ERC), Health Holland (HH), Mendus, BioNovion, Aduro Biotech, Vicinivax, Genmab, and IMMIOS (all paid to the institute); received non-financial support from BioNTech, Surflay Nanotec, and Merck Sharp \u0026amp; Dohme; is stock option holder in Sairopa; all unrelated to the submitted work.\u003c/p\u003e"},{"header":"REFERENCES","content":"\u003col\u003e\n\u003cli\u003eHaslam A, Kim MS, Prasad V (2021) Updated estimates of eligibility for and response to genome-targeted oncology drugs among US cancer patients, 2006-2020. Ann Oncol 32:926-32. doi: 10.1016/j.annonc.2021.04.003\u003c/li\u003e\n\u003cli\u003eYamaguchi H, Hsu JM, Sun L, et al (2024) Advances and prospects of biomarkers for immune checkpoint inhibitors. Cell Rep Med 5:101621. doi: 10.1016/j.xcrm.2024.101621\u003c/li\u003e\n\u003cli\u003eWang X, Muzaffar J, Kirtane K, et al (2022) T cell repertoire in peripheral blood as a potential biomarker for predicting response to concurrent cetuximab and nivolumab in head and neck squamous cell carcinoma. J Immunother Cancer 10: e004512. doi: 10.1136/jitc-2022-004512\u003c/li\u003e\n\u003cli\u003eNaranbhai V, Viard M, Dean M, et al (2022) HLA-A*03 and response to immune checkpoint blockade in cancer: an epidemiological biomarker study. Lancet Oncol 23:172-84. doi: 10.1016/S1470-2045(21)00582-9\u003c/li\u003e\n\u003cli\u003eAiba T, Hattori C, Sugisaka J, et al (2021) Gene expression signatures as candidate biomarkers of response to PD-1 blockade in non-small cell lung cancers. PLoS One 16:e0260500. doi: 10.1371/journal.pone.0260500\u003c/li\u003e\n\u003cli\u003evan Wilpe S, Wosika V, Ciarloni L, et al (2021) Whole blood transcriptome profiling identifies DNA replication and cell cycle regulation as early marker of response to anti-PD-1 in patients with urothelial cancer. Cancers (Basel) 13:4660. doi: 10.3390/cancers13184660\u003c/li\u003e\n\u003cli\u003eNagumo Y, Kandori S, Kojima T, et al (2022) Whole blood gene expression profiles correlate with response to immune checkpoint inhibitors in patients with metastatic renal cell carcinoma. Cancers (Basel) 14:6207. doi: 10.3390/cancers14246207\u003c/li\u003e\n\u003cli\u003eFriedlander P, Wassmann K, Christenfeld AM, et al (2017) Whole blood RNA transcript-based models can predict clinical response in two large independent clinical studies of patients with advanced melanoma treated with the checkpoint inhibitor, tremelimumab. J Immunother Cancer 2017;5:67. doi: 10.1186/s40425-017-0272-z\u003c/li\u003e\n\u003cli\u003eBhattacharya A, Bense RD, Urz\u0026uacute;a-Traslavi\u0026ntilde;a CG, et al (2020) Transcriptional effects of copy number alterations in a large set of human cancers. Nat Commun 11:715. doi: 10.1038/s41467-020-14605-5\u003c/li\u003e\n\u003cli\u003eKnapen DG, Hone Lopez S, de Groot DJA, et al (2024) Independent transcriptional patterns reveal biological processes associated with disease-free survival in early colorectal cancer. Commun Med (Lond). 4:79. doi: 10.1038/s43856-024-00504-z\u003c/li\u003e\n\u003cli\u003eLachmann A, Torre D, Keenan AB, et al (2018) Massive mining of publicly available RNA-seq data from human and mouse. Nat Commun 9:1366. doi: 10.1038/s41467-018-03751-6\u003c/li\u003e\n\u003cli\u003eLove MI, Huber W, Anders S (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. doi: 10.1186/s13059-014-0550-8\u003c/li\u003e\n\u003cli\u003eLiberzon A, Birger C, Thorvaldsd\u0026oacute;ttir H, Ghandi M, Mesirov JP, Tamayo P (2015) The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1:417-25. doi: 10.1016/j.cels.2015.12.004\u003c/li\u003e\n\u003cli\u003eKairov U, Cantini L, Greco A, et al (2017) Determining the optimal number of independent components for reproducible transcriptomic data analysis. BMC Genomics 18:712. doi: 10.1186/s12864-017-4112-9\u003c/li\u003e\n\u003cli\u003eChung N, Zhang XD, Kreamer A, et al (2008) Median absolute deviation to improve hit selection for genome-scale RNAi screens. J Biomol Screen 13:149-58. doi: 10.1177/1087057107312035\u003c/li\u003e\n\u003cli\u003eDobin A, Davis CA, Schlesinger F, et al (2013) STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29:15-21. doi: 10.1093/bioinformatics/bts635\u003c/li\u003e\n\u003cli\u003eVan der Auwera GA, Carneiro MO, Hartl C, et al (2013) From FastQ data to high‐confidence variant calls: the genome analysis toolkit best practices pipeline. Curr Protoc Bioinformatics 43:1.10.1-11.10.33. doi: 10.1002/0471250953.bi1110s43\u003c/li\u003e\n\u003cli\u003eOrenbuch R, Filip I, Comito D, Shaman J, Pe\u0026apos;er I, Rabadan R (2020) arcasHLA: high-resolution HLA typing from RNAseq. Bioinformatics 36:33-40. doi: 10.1093/bioinformatics/btz474\u003c/li\u003e\n\u003cli\u003eLing YH, Chen Y, Leung KN, Chan KM, Liu WK (2023) Cell cycle regulation of the psoriasis associated gene CCHCR1 by transcription factor E2F1 PloS One 18:e0294661. doi: 10.1371/journal.pone.0294661\u003c/li\u003e\n\u003cli\u003eShifrut E, Carnevale J, Tobin V, et al (2018) Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell 2018;175:1958-1971.e1 doi: 10.1016/j.cell.2018.10.024\u003c/li\u003e\n\u003cli\u003eCarnevale J, Shifrut E, Kale N, et al (2022) RASA2 ablation in T cells boosts antigen sensitivity and long-term function. Nature 609:174-82. doi: 10.1038/s41586-022-05126-w\u003c/li\u003e\n\u003cli\u003eKim KH, Cho J, Ku BM, et al (2019) The first-week proliferative response of peripheral blood PD-1+CD8+ T cells predicts the response to anti-PD-1 therapy in solid tumors. Clin Cancer Res 25:2144-54. doi: 10.1158/1078-0432.CCR-18-1449\u003c/li\u003e\n\u003cli\u003eKamphorst AO, Pillai RN, Yang S, et al (2017) Proliferation of PD-1+ CD8 T cells in peripheral blood after PD-1-targeted therapy in lung cancer patients. Proc Natl Acad Sci U S A 114:4993-98. doi: 10.1073/pnas.1705327114\u003c/li\u003e\n\u003cli\u003eZhou F, Shen C, Xu J, et al (2016) Epigenome-wide association data implicates DNA methylation-mediated genetic risk in psoriasis. Clin Epigenetics 8:131. doi: 10.1186/s13148-016-0297\u003c/li\u003e\n\u003cli\u003ePrieto-P\u0026eacute;rez R, Solano-L\u0026oacute;pez G, Cabaleiro T, et al (2015) Polymorphisms associated with age at onset in patients with moderate-to-severe plaque psoriasis. J Immunol Res 2015:101879. doi: 10.1155/2015/101879\u003c/li\u003e\n\u003cli\u003eCheung YH, Watkinson J, Anastassiou D (2011) Conditional meta-analysis stratifying on detailed HLA genotypes identifies a novel type 1 diabetes locus around TCF19 in the MHC. Hum Genet 129:161-76. doi: 10.1007/s00439-010-0908-2\u003c/li\u003e\n\u003cli\u003eSugawara T, Shimizu H, Hoshi N, Nakajima A, Fujimoto S (2003) Steroidogenic acute regulatory protein-binding protein cloned by a yeast two-hybrid system. J Biol Chem 278:42487\u0026ndash;94. doi: 10.1074/jbc.M302291200\u003c/li\u003e\n\u003cli\u003eTiala I, Suomela S, Huuhtanen J, et al (2020) The CCHCR1 (HCR) gene is relevant for skin steroidogenesis and downregulated in cultured psoriatic keratinocytes. J Mol Med 85:589\u0026ndash;601. doi: 10.1007/s00109-006-0155-0\u003c/li\u003e\n\u003cli\u003eMahata B, Pramanik J, van der Weyden L, et al (2020) Tumors induce de novo steroid biosynthesis in T cells to evade immunity. Nat Comm 11:3588. doi: 10.1038/s41467-020-17339-6\u003c/li\u003e\n\u003cli\u003evon Hoff L, K\u0026auml;rgel E, Franke V, et al (2019) Autocrine LTA signaling drives NF-\u0026kappa;B and JAK-STAT activity and myeloid gene expression in Hodgkin lymphoma. Blood 133:1489-94. doi: 10.1182/blood-2018-08-871293\u003c/li\u003e\n\u003cli\u003eSchmidt R, Steinhart Z, Layeghi M, et al (2022) CRISPR activation and interference screens decode stimulation responses in primary human T cells. Science 375:eabj4008. doi: 10.1126/science.abj4008\u003c/li\u003e\n\u003cli\u003eCheng Y, Liu Y, Xu D, et al (2024) An engineered TNFR1-selective human lymphotoxin-alpha mutant delivered by an oncolytic adenovirus for tumor immunotherapy. Biochim Biophys Acta Mol Basis Dis 1870:167122. doi: 10.1016/j.bbadis.2024.167122\u003c/li\u003e\n\u003cli\u003eBeldi G, Khosravi M, Abdelgawad ME, et al (2020) TNF\u0026alpha;/TNFR2 signaling pathway: an active immune checkpoint for mesenchymal stem cell immunoregulatory function. Stem Cell Res Ther 11:281. doi: 10.1186/s13287-020-01740-5\u003c/li\u003e\n\u003cli\u003eLee HS, Sun H, Lagarrigue F, et al (2022) Phostensin enables lymphocyte integrin activation and population of peripheral lymphoid organs. J Exp Med 219:e20211637. doi: 10.1084/jem.20211637\u003c/li\u003e\n\u003cli\u003eArmstrong DL, Zidovetzki R, Alarc\u0026oacute;n-Riquelme ME, et al (2014) GWAS identifies novel SLE susceptibility genes and explains the association of the HLA region. Genes Immun 15:347-54. doi: 10.1038/gene.2014.23\u003c/li\u003e\n\u003cli\u003eHakonarson H, Grant SF, Bradfield JP, et al (2007) A genome-wide association study identifies KIAA0350 as a type 1 diabetes gene. Nature 448:591-94. doi: 10.1038/nature06010\u003c/li\u003e\n\u003cli\u003eMatzaraki V, Kumar V, Wijmenga C, Zhernakova A (2017) The MHC locus and genetic susceptibility to autoimmune and infectious diseases. Genome Biol 18:76. doi: 10.1186/s13059-017-1207-1\u003c/li\u003e\n\u003cli\u003eKulski JK, Suzuki S, Shiina T (2022) Human leukocyte antigen super-locus: nexus of genomic supergenes, SNPs, indels, transcripts, and haplotypes. Hum Genome Var 9:49. doi: 10.1038/s41439-022-00226-5\u003c/li\u003e\n\u003cli\u003eGambino CM, Aiello A, Accardi G, Caruso C, Candore G (2018) Autoimmune diseases and 8.1 ancestral haplotype: An update. HLA 92:137-43. doi: 10.1111/tan.13305\u003c/li\u003e\n\u003cli\u003eRozeman EA, Hoefsmit EP, Reijers ILM, et al (2021) Survival and biomarker analyses from the OpACIN-neo and OpACIN neoadjuvant immunotherapy trials in stage III melanoma. Nat Med 27:256-63. doi: 10.1038/s41591-020-01211-7\u003c/li\u003e\n\u003cli\u003eJou E (2023) Type 1 and type 2 cytokine-mediated immune orchestration in the tumour microenvironment and their therapeutic potential. Explor Target Antitumor Ther 4:474-97. doi: 10.37349/etat.2023.00146\u003c/li\u003e\n\u003cli\u003eWynn T (2015) Type 2 cytokines: mechanisms and therapeutic strategies. Nat Rev Immunol 15:271\u0026ndash;82. doi: 10.1038/nri3831\u003c/li\u003e\n\u003cli\u003eGao X, Wang X, Yang Q, et al (2015) Tumoral expression of IL-33 inhibits tumor growth and modifies the tumor microenvironment through CD8+ T and NK cells. J Immunol 194:438-45. doi: 10.4049/jimmunol.1401344\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 3 are available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"cancer-immunology-immunotherapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ciim","sideBox":"Learn more about [Cancer Immunology, Immunotherapy](http://link.springer.com/journal/262)","snPcode":"262","submissionUrl":"https://submission.nature.com/new-submission/262/3","title":"Cancer Immunology, Immunotherapy","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Immunotherapy, Immune Checkpoint Inhibitors, Biomarker, Gene Expression Profiling, Transcriptomics ","lastPublishedDoi":"10.21203/rs.3.rs-6843569/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6843569/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWhole blood (WB) transcriptomics offers a minimal-invasive method to assess patients’ immune system. This study aimed to identify transcriptional patterns in WB associated with clinical outcomes in patients treated with immune checkpoint inhibitors (ICIs).\u003c/p\u003e\n\u003cp\u003eWe performed RNA-sequencing on pre-treatment WB samples from 145 patients with advanced cancer. Additionally, we compiled a separate dataset of 14,085 WB transcriptomes from diverse health backgrounds from public repositories and applied consensus-independent component analysis (c-ICA) to identify transcriptional components (TCs). The biological processes represented by these TCs were elucidated using gene set enrichment analysis. The activity of the TCs was then quantified in the 145 WB profiles and analyzed for associations with tumor response, progression-free survival, and overall survival using univariate and multivariate analyses in a permutation framework. RNA-sequencing variant calling was performed, and the activity of the TCs was assessed in specific cell lineages using a single-cell immune cell atlas of the human hematopoietic system.\u003c/p\u003e\n\u003cp\u003ec-ICA on 14,085 WB transcriptomes identified 1,262 distinct TCs representing various cellular processes. Of these, 18 TCs were associated with ≥1 outcome parameter, with three specifically linked to tumor response. Top genes in these three TCs included CCHCR1, TCF19, LTA, DDX39B, and PPPR1R18. RNA-sequencing variant calling and single-cell transcriptome projections revealed associations between these four TCs and germline variants.\u003c/p\u003e\n\u003cp\u003eThese findings support the potential of the identified WB-based transcriptional patterns to complement tumor characteristics in predictive and prognostic models for improved patient stratification.\u003c/p\u003e","manuscriptTitle":"Upfront whole blood transcriptional patterns in patients receiving immune checkpoint inhibitors associate with clinical outcome","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-13 05:45:03","doi":"10.21203/rs.3.rs-6843569/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-06-25T06:03:31+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-23T07:31:21+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-22T06:26:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"124233774148157617483112495558155488600","date":"2025-06-12T15:04:37+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"11235916069249902991422021968878945645","date":"2025-06-11T02:30:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"136705891903206949419227104509041871616","date":"2025-06-10T15:54:50+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-10T09:53:06+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-09T13:58:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-09T13:55:11+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cancer Immunology, Immunotherapy","date":"2025-06-07T14:59:14+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cancer-immunology-immunotherapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ciim","sideBox":"Learn more about [Cancer Immunology, Immunotherapy](http://link.springer.com/journal/262)","snPcode":"262","submissionUrl":"https://submission.nature.com/new-submission/262/3","title":"Cancer Immunology, Immunotherapy","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"cdd4d20b-7bdc-4bb6-b5b6-f980f0b9ca64","owner":[],"postedDate":"June 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-09-15T15:58:28+00:00","versionOfRecord":{"articleIdentity":"rs-6843569","link":"https://doi.org/10.1007/s00262-025-04155-4","journal":{"identity":"cancer-immunology-immunotherapy","isVorOnly":false,"title":"Cancer Immunology, Immunotherapy"},"publishedOn":"2025-09-11 15:57:01","publishedOnDateReadable":"September 11th, 2025"},"versionCreatedAt":"2025-06-13 05:45:03","video":"","vorDoi":"10.1007/s00262-025-04155-4","vorDoiUrl":"https://doi.org/10.1007/s00262-025-04155-4","workflowStages":[]},"version":"v1","identity":"rs-6843569","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6843569","identity":"rs-6843569","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}