Matched Tissues Research Articles

Growth regulated oncogene-α contribute to EMT/MMPs pathway by binding its receptors in head and neck squamous cell carcinoma

Squamous cell carcinoma (SCC) is the most common malignant tumor of the head and neck and generally detected in the late stages when the cancer has advanced, and therefore has a poor prognosis and survival rate. A high expression of growth-related oncogene alpha (Groα) is associated with tumor metastasis and invasion and the poor survival rate of patients. Microarray reveals that Groα exhibits a cancer-specific response in HNSCC. Quantitative real-time PCR (qRT-PCR) results concerning the mRNA expression of Groα in HNSCC tissues; indicate that Groα was more highly expressed in HNSCC than in non-cancerous matched tissue (NCMT). The serum of HNSCC patients and healthy subjects demonstrates that the expression of Groα in the HNSCC patients significantly exceeded than in healthy subjects. Furthermore, exposure Groα to stimulated the proliferation, clonogenicity and migration with HNSCC cells (SCC4, SCC9, SCC25 and OECM-1), yielding a stronger response than in non-malignant HaCaT and DOK cells. A high expression of Groα and its receptors CXCR1/2 (chemokine (C-X-C motif) receptor) in HNSCC tissues are highly correlated with tumor progression stage and metastasis. Following the treatment of SCC25 and OECM-1 cells with Groα, β-catenin, matrix metalloproteinases (MMP)-2, MMP-7 and MMP-9 expressions significantly increased but E-cadherin expression was slightly decreased, suggesting that the EMT and metastasis processes were activated by Groα. These findings constitute the first evidence that Groα promotes epithelial mesenchymal transition (EMT) and MMPs expressions in HNSCC via activating CXCR1/2, suggesting a role for Groα in mediating metastasis and its potential as a therapeutic target.

Comprehensive characterization reveals sputum supernatant as a valuable alternative liquid biopsy for genome profiling in advanced non-small cell lung cancer

BackgroundSputum biopsies offer unique advantages such as non-invasiveness and convenient collection. The one investigation so far on sputum for genome profiling in advanced non-small cell lung cancer (aNSCLC) suggested promising performance. However, it remains undefined whether clinicohistologic characteristics were associated with performance and how this knowledge could help guide choice of liquid biopsy.MethodsTargeted sequencing with a 520-gene panel was performed on prospectively collected matched tumor tissue (TIS), plasma (PLA), and sputum supernatant (SPU) from 71 aNSCLC patients (NCT05034445). Genomic alteration detection was characterized in a series of aspects and interrogated for association with 14 clinicohistologic features. Nomograms were constructed with logistic regression for predicting the liquid biopsy type with greater sensitivity.ResultsCompared with PLA, SPU showed comparable quality control metrics, mutation detection rate (SPU: 67.6%, PLA: 70.4%), concordance with tumor tissue (67.6% vs. 73.2%), and correlation with tissue-based tumor mutation burden levels (r = 0.92 vs. 0.94). For driver alterations, detection was less sensitive with SPU (50.0%) than PLA (63.5%) in the entire cohort but similarly or more sensitive in patients with centrally located lung tumors or smoking history or for altered ALK or KRAS. Two nomograms were constructed and enabled predicting the probability of superior sensitivity with SPU with moderate to borderline high accuracy.ConclusionIn addition to demonstrating comparable performance in multiple aspects, this study is the first to propose nomograms for choosing liquid biopsy based on clinicohistologic characteristics. Future research is warranted to delineate the clinical utility of sputum for genome profiling.

Markers of Immune Cell Exhaustion as Predictor of Survival in Surgically-Treated Early-Stage NSCLC.

BackgroundTumor tissue as well as regional lymph nodes are removed during curative surgery for early-stage non-small cell lung cancer (NSCLC). These tissues provide a unique snapshot of the immune cell composition at the time of surgery. We investigated the immune landscape in matched tumor tissue, tumor bearing (tb) and non-tumor bearing (ntb) N1 as well as N2 lymph nodes (LNs) in patients with NSCLC and its relation to survival.MethodsInternal hospital databases were screened for surgically treated NSCLC patients for whom tumor tissue, tbLNs as well as N1 and N2 ntbLNs were available. Clinical as well as demographic data were extracted from hospital records. Expression profiling of 770 immune-related genes was performed using the PanCancer IO 360 panel by NanoString Technologies.ResultsWe identified 190 surgically treated patients of whom 16 fulfilled inclusion criteria and had sufficient archived tissue. The Tumor Immune Dysfunction and Exclusion (TIDE) score in N1 tumor-free lymph nodes was associated with OS. TIM-3 expression was inversely correlated with TIDE scores in affected LNs, N1 and N2 ntbLNs. Levels of CD8 expression were significantly higher in TIDE High compared to TIDE Low patients. TIM-3 and PD-L1 were selected for the final model for OS in multivariate regression in more than one tissue.ConclusionLevels of immune cell exhaustion markers may indicate a dysfunctional immune status and are associated with survival after curative surgery in NSCLC.

Immune Dysfunction Mediated by the ceRNA Regulatory Network in Human Placenta Tissue of Intrahepatic Cholestasis Pregnancy.

Pregnancy-related intrahepatic cholestasis (ICP) is a serious complication with adverse perinatal outcomes of preterm labor, fetal distress, or stillbirth. As a result, it is important to investigate and identify the potential critical pathogenic mechanisms of ICP. First, we collected the placental tissues from the ICP with placental weight and fetal birth weight loss for the whole transcriptome sequencing. Then we analyzed the differentially expressed (DE) circRNAs (DEcircRNAs) by SRPBM, DElncRNAs by FRKM, DEmiRNAs by TPM, and DEmRNAs by TPM and RSEM. Based on differential expression of term pregnancy placental tissues from pregnancies impacted by ICP (n=7) as compared to gestational aged matched control tissues (n=5), the circ/lncRNA-miRNA-mRNA competitive endogenous RNA (ceRNA) regulatory networks were constructed. The ceRNA regulatory networks covered 3,714 events, including 21 DEmiRNAs, 36 DEcircRNAs, 146 DElncRNAs, and 169 DEmRNAs. According to the functional analysis, ICP complications were linked to the immune system, signal transduction, endocrine system, cell growth and death, and transport and catabolism. Further evidence suggested that the expression of immune-related genes KLRD1, BRAF, and NFATC4 might have a potential ceRNA mechanism by individual lncRNA sponging miR372-3p, miR-371a-3p, miR-7851-3p, and miR-449a to control downstream the level of TNF-α, IFN-γ, and IL-10, thereby regulating the pathophysiology of ICP. Furthermore, our results were validated by the qRT-PCR, western blotting and ELISA assays. In conclusion, this study is the first to evaluate placental ceRNA networks in pregnancies affected by ICP, showing alterations in immune regulatory networks which may impact fetal and placental growth. Overall our these data suggest that the ceRNA regulatory network may refine biomarker predictions for developing novel therapeutic approaches in ICP.

Abstract 5105: Comparing comprehensive NGS panel(s) based approach versus tissue whole exome sequencing (WES) using matched tissue and circulating tumor DNA from late-stage colorectal cancer patients

Abstract Background: The ctDNA shedding rate, concentration and stability are associated with tumor type, stage, ctDNA degradation and clearance mechanisms. Tumoral heterogeneity together with tissue of origin related genomic alterations adds complexity to genomic profiling. The value of NGS panels capable of detecting variants in both tissue and ctDNA are well recognized. Therefore, it is essential to assess the comprehensive NGS panel(s) performance in both liquid biopsy and tissue by comparing to tissue WES. In this study, we systematically conducted performance evaluation of llumina TruSight™ Oncology 500 gene (TSO500) NGS assay(s) using matched tissue and ctDNA specimens Methods: Matched tissue and plasma samples from seven treatment naive stage III/IV CRC patients were utilized for Illumina TSO500 sequencing and WES. Tissue samples were separately analyzed by both WES and the TSO500 tissue panel while extracted cfDNA was analyzed by TSO500 ctDNA panel. Illumina BaseSpace TruSight Oncology 500 platform (CombinedVariantOutput.tsv) was applied to select non-synonymous (frameshift, in-frame, nonsense, missense, translation start site, nonstop and splice site) variants in both cfDNA and tissue samples. Genes found to be mutated by tissue WES were filtered for corresponding mutated genes detected by TSO500 analysis. Next, using maftools, we characterize the most frequently mutated genes and the mutations observed for each patient sample. Results: 1,300 variants and 33 copy number variants were observed in the liquid biopsy samples whereas 1,199 variants, 9 genes with high copy number variations and 3 gene fusions were observed in the TSO500 tissue panel. Of these mutations, 889 variants (56.9%) were common between the liquid and tissue in the TSO500 panels. We observed a significant correlation across WES and TSO500-called variants (TSO500 tissue/TSO500 cfDNA r2=0.81, WES/TSO500 tissue r2=0.91 (47 variants), WES/TSO500 cfDNA r2=0.84 (17 variants). In the TSO tissue and cfDNA datasets, 19/20 most commonly mutated genes were the same, while FAT1 was the most commonly mutated gene in both tissue and ctDNA samples. In addition, only 2/20 (APC and TP53) most commonly mutated genes found in the tissue WES dataset were included in the top 20 for two TSO500 datasets. Furthermore, the RTK-RAS pathway had mutations in 22 genes in the liquid and tissue TSO500, by far the most of any pathway. Conclusions: Significant concordance between the TSO500 liquid biopsy and tissue datasets were demonstrated, but lesser concordance with the tissue WES dataset. The evidence supports the potential of leveraging both ctDNA and FFPE based genomic observation to enable precision oncology. A deeper analytical dive into each dataset is on-going. Citation Format: Dennis O'Rourke, Roman Alpatov, Danyi Wang, Zheng Feng, Juergen Scheuenpflug. Comparing comprehensive NGS panel(s) based approach versus tissue whole exome sequencing (WES) using matched tissue and circulating tumor DNA from late-stage colorectal cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5105.

Abstract 611: Assessment of tumor mutation burden and microsatellite instability by single-cell circulating tumor cell genomic profiling

Abstract The success of immune checkpoint inhibitors rests on biomarkers such as tumor mutation burden (TMB) and microsatellite instability (MSI), both FDA-approved predictors of anti PD-1/L1 therapy benefit. Tissue biopsies often collected once in the metastatic setting through an invasive procedure, or archived primary tumor tissue often collected much prior to treatment consideration, are the specimen types of choice for biomarker identification. The tissue sample originates from a limited region of one disease site, which may limit its usefulness given intra-patient tumor heterogeneity. TMB and MSI measurement by liquid biopsy, including proteins, circulating tumor cells (CTC), and cell-free circulating tumor DNA (ctDNA), is an attractive, minimally-invasive way to obtain a real-time picture of the entire disease. While TMB and MSI assessment from ctDNA have been reported, their measurement can be limited by low ctDNA tumor fraction. Single-cell next generation sequencing of CTC, on the other hand, is a particularly well-suited, but largely unexplored method of measuring TMB and MSI to complement tissue and ctDNA for better overall specificity of detection. In this proof-of-concept study, we show the ability to detect single-cell TMB and MSI. We analyzed 14 CTC and 4 ctDNA samples from 6 metastatic breast cancer patients, as well as 3 single cells and 1 cell pellet sample each from HCT-116 (MSI-High) and WiDr (MSI-Low) cell lines. CTC and cell line cells were enriched with the CellSearch® system and/or isolated with the DEPArray™ system. Whole genome amplified single-cell DNA was sequenced with the Oncomine Comprehensive Assay covering ~500 genes and 1.1Mb of genomic space. TMB and MSI scores obtained in CTC and ctDNA were compared to those measured in matched clinical tissue biopsies. Single-cell TMB scores and MSI status were assessable in all CTC tested. CTC TMB scores were highly concordant with the matched tissue samples (r=1.00), as were ctDNA TMB scores (r=0.98) in patients with assessable TMB scores in both biospecimen types compared. Importantly, TMB was detectable in CTC from one patient whose tissue sample was inadequate for clinical sequencing, and from another patient with inadequate, low tumor fraction ctDNA. Intriguingly, one patient harbored 3 TMB-high and 2 TMB-low CTC, potentially indicating intra-patient TMB heterogeneity. The known MSI-low status from clinical tumor tissue sequencing was correctly detected in CTC and ctDNA from all patients. MSI status and scores from single cells of HCT-116 and WiDr cell lines purified with the DEPArray™ system (mimicking CTC isolation), perfectly matched that of the corresponding cell pellet samples (r=1.00). Taken together, these data suggest the potential validity and continued interrogation of potential utility of CTC TMB and MSI detection to complement tissue and ctDNA in guiding checkpoint inhibitor immunotherapy. Citation Format: Andi K. Cani, Emily M. Dolce, Chia-Jen Liu, Brittany Rupp, Elizabeth P. Darga, Costanza Paoletti, Dafydd G. Thomas, Yi-Mi Wu, Dan R. Robinson, Sunitha Nagrath, Arul M. Chinnaiyan, Scott A. Tomlins, Aaron M. Udager, John M. Carethers, Erin F. Cobain, Daniel F. Hayes. Assessment of tumor mutation burden and microsatellite instability by single-cell circulating tumor cell genomic profiling [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 611.

Abstract 1374: Discovery of MHC class I and class II neoantigens in lung cancer in needle biopsy tissue samples using an optimized high-throughput workflow

Abstract Human leukocyte antigen-associated peptides, known as immunopeptides, play an essential role in adaptive immunity by activating and ensuring the specificity of T-cells. The identification and quantification of the immunopeptidome bear the potential to enable personalized treatments, especially in cancers, vaccines, infectious, and autoimmune diseases. Mass spectrometry is currently the only technology that can reliably measure and identify immunopeptide profiles of biological samples on a large scale. However, the usually high sample input amount and poor scalability are limiting. Here, we introduce a semi-automated workflow to robustly identify immunopeptides from low amounts of cultured cells and tissue samples by systematically optimizing each step of the sample preparation and acquisition. We optimized the native lysis and immunoprecipitation workflow while ensuring scalability and reproducibility. Leveraging the magnetic properties of the beads, 1,000 samples can be processed within a week by a single operator. The established sample preparation offers high reproducibility and identifications of good quality. For class-I immunopeptides, &gt60% of the peptides identified are 9-mers, &gt80% predicted strong binders, and the expected amino acids are enriched at the anchor positions. For class-II, &gt50% of the peptides identified are 14-to-16-mers, and &gt50% are predicted strong binders. Furthermore, the pipeline is highly sensitive as we could still identify over 2,800 class-I immunopeptides when processing as little as 2.5 mg fresh frozen tissue and &gt9,000 class-I and &gt12,000 class-II immunopeptides when preparing 10 million JY cells. Overall, the pipeline is scalable, highly reproducible, and results in high-quality identifications while supporting very limited sample input. Finally, we measured a cohort of 12 cancerous and matched healthy lung tissues from as little as 15 mg tissue, whereby we could identify &gt11,000 class-I immunopeptides and &gt9,000 class-II on average. For class-I, matched samples clustered together, while &gt3,000 immunopeptides were upregulated in the cancer tissues, with a significant enrichment for proteins related to lung cancer. Overall, we established a scalable, efficient pipeline for cell line and tissues immunopeptidomics for class-I and II that generates high-quality identifications and that only requires small amounts of input material and is ready to shed light into immunopeptidomics heterogeneity through large-scale profiling of patients. Citation Format: Ilja E. Shapiro, Luca Raess, Marco Tognetti, Tikira Temu, Oliver M. Bernhardt, Yuehan Feng, Roland Bruderer, Lukas Reiter. Discovery of MHC class I and class II neoantigens in lung cancer in needle biopsy tissue samples using an optimized high-throughput workflow [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1374.

Abstract 1571: Cancer associated fibroblasts sustain critical dependency of pancreatic cancer cells on exogenous lipids

Abstract Pancreatic ductal adenocarcinoma (PDAC) cells derive their resistance to therapy and aggressive clinical course from the symbiotic signaling and metabolic interactions with cancer-associated fibroblasts (CAFs). CAFs have been shown to provide nutrients to "parasitic" PDAC cells including water soluble glucose and amino acids. In hypoxic tumor microenvironments, aggressive PDAC are functional auxotroph for lipids, and scavenge exogenous lipids to build cellular membranes. However, the mechanism by which PDAC cells obtain these water-insoluble essential membrane building materials remains poorly understood. To gain insights into the mechanism of lipid scavenging, we generated cholesterol-auxotrophic human and mouse PDAC PDAC cells since cholesterol is the major constituent of lipid cellular membranes. Here, we discovered that PDAC cells utilize CAFs as a main source of lipids in vivo through direct heterotypic cellular contacts. In this process, PDAC cells directly acquire the CAF plasma membrane (PM) via trogocytosis. In response to yet unidentified paracrine “feed me” signals activating CAF-dependent trogocytosis, CAFs upregulate phosphatidylserine (PtdSer) on the outer leaflet of the PM. Consequently, blockade of PtdSer on CAFs partially deterred trogocytic transfer of CAF membranes to PDAC cells in vitro. Furthermore, Ca2+-dependent phospholipid scramblase TMEM16F is a critical regulator of PtdSer externalization. TMEM16F is highly expressed in human PDAC CAFs compared to fibroblasts isolated from the matching adjacent non-malignant pancreatic tissues CAFs deficient in TMEM16F scramblase exhibited markedly reduced ability to support the growth of cholesterol-auxotrophic PDAC cells cultured in lipid depleted media. We propose trogocytosis as a new mode of lipid scavenging by PDAC cells from CAFs involving activation of Ca2+-dependent phospholipid scramblase TMEM16F in CAFs and increased expression of PtdSer as “eat me” signals on CAF PM. To inactivate trogocytosis, we nominate TMEM16F as a plausible PDAC therapy target using clinically available TMEM16 inhibitors with a potential for impact on treatment of PDAC patients in the near term. Citation Format: Charline Ogier, Alena Klochkova, Battuya Bayarmagnai, Linara Gabitova, Diana Restifo, Aizhan Surumbayeva, Janusz Franco-Barraza, Debora Vendramini Costa, Ralph Francescone, Jaye Gardiner, Emmanuelle Nicolas, Elizabeth A. Handorf, Kathy Q. Cai, Edna Cukierman, Igor Astsaturov. Cancer associated fibroblasts sustain critical dependency of pancreatic cancer cells on exogenous lipids [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1571.

Abstract 3074: (Phospho)proteomics profiling of patient-derived colon cancer organoids for the discovery of putative drug targets

Abstract Introduction: Colorectal cancer is the second cause of cancer related deaths worldwide, with the poor survival outcomes attributable to the presence of metastases. To tackle this problem, (phospho)proteomics analysis enables valuable insights into changes of protein expression and signalling in cancer that can be perturbed by drugs. To study mechanisms driving metastasis and perform subsequent drug testing, patient derived organoids (PDOs) are in development as preclinical models. PDOs are obtained by 3D culture of tumour tissue ex-vivo, which enables them to retain the heterogeneity and architecture of the tumor source. To identify putative drug targets for colon cancer metastasis, we performed comparative (phospho)proteomics analysis of metastatic colon tissues and their matched PDOs. Methods: (Phospho)proteomics profiling was performed on PDOs generated from 10 patients with colon metastases to the liver, 10 matched tumour tissues and 4 normal colon mucosa. Unsupervised analysis of the (phospho)proteomic landscape was used to identify differentially expressed genes and pathways in colon metastases and their matched PDOs, compared to normal colon mucosa. We focused on phosphosites, proteins and pathways showing consistently altered expression in tumour tissues and PDOs. Kinase-substrate enrichment analysis was used to identify aberrantly activated kinases, based on the abundance of phosphorylation on the substrates. Putative drugs were selected using the consensus expression response to multiple small molecule drugs across cell lines and conditions from the Library of Integrated Network-Based Cellular Signatures (LINCS) database. Drug treatment of drug candidates was performed using the MTT Cell Proliferation Assay. Results: Using the approach described above we identified 103 differentially expressed proteins and 236 phosphosites between tumour tissues and normals, which were also detected in PDOs, with overall high correlation of t statistics (0.6 Spearman’s rho) between tumor tissues and PDOs. MYC-targets, G2M checkpoints and E2F-targets were amongst the top positively enriched pathways in common, and the LINCS analysis identified multi-kinase inhibitor Nintedanib and NFKB pathway-targeting KIN0-260 as putative drugs for upregulated proteins. The kinase CSNK2A1 was overactivated in both groups, pointing towards the use of CK2 inhibitors for drug testing. Treatment at 5 days with Nintedanib and KIN001-260 on the PDOs resulted in significant reduction of cell viability compared with 5-FU. Conclusion: Our study provides evidence that PDOs recapitulate relevant tumor features at the proteomic and phosphoproteomic levels, supporting the utility of this ex-vivo model as a tool for drug sensitivity testing for metastatic colon cancer. Citation Format: Gina Faye Boot, Federica Panebianco, John Gallon, Charlotte Kiu Yan Ng, Salvatore Piscuoglio. (Phospho)proteomics profiling of patient-derived colon cancer organoids for the discovery of putative drug targets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3074.

Abstract 5136: Development of a novel gene expression signature that correlates with intratumor microvessel density and antitumor activity of lenvatinib

Abstract Lenvatinib is a multiple receptor tyrosine kinase inhibitor and an angiogenesis inhibitor targeting mainly vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF) receptors. In our previous study, we reported that human xenograft tumors sensitive to lenvatinib showed higher microvessel density (“MVD (IHC)”) than relatively resistant tumors in immunohistochemistry (IHC). Although MVD (IHC) is a potential indicator to predict antitumor activity of lenvatinib, available IHC data are limited. Here, we investigated a novel gene expression signature (MVD gene score) calculated from RNA-sequencing (RNA-seq) data that reflects MVD (IHC) and antitumor effect of lenvatinib. Tumor tissues were collected from 12 mouse syngeneic tumor models followed by RNA-seq. We also stained CD31 in the same tumors and calculated MVD (IHC) score by dividing the number of CD31 positive blood vessels by tumor area. Genes that met the following criteria were selected for MVD gene score calculation: (1) genes that are specifically expressed on endothelial cell cluster in in-house single cell RNA-seq data using Hepa 1-6 mouse hepatocellular carcinoma tumor tissues; (2) well-known angiogenesis markers from literature or genes whose expression show significant correlation with MVD (IHC) in mouse syngeneic tumor models. MVD gene score was calculated as the average of logarithm transformed gene expression values. In order to examine applicability of MVD gene score, we compared MVD (IHC) and MVD gene score to lenvatinib antitumor activities on mouse syngeneic tumors. We measured MVD (IHC) and computed MVD gene score from RNA-seq data using matched tumor tissues of 12 mouse syngeneic models. MVD gene score was correlated with MVD (IHC). As reported, lenvatinib sensitive tumor models showed higher MVD (IHC) score. They also showed higher MVD gene score calculated from RNA-seq data. Our newly developed MVD gene score has a potential to predict the antitumor activity of angiogenic inhibitors such as lenvatinib using RNA-seq data in tumor samples. Further validation study is ongoing using various human tumor samples. Citation Format: Megumi Kuronishi, Yoichi Ozawa, Takayuki Kimura, Yu Kato. Development of a novel gene expression signature that correlates with intratumor microvessel density and antitumor activity of lenvatinib [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5136.

Abstract 1508: Chromatin accessibility of primary human cancers ties regional mutational processes with tissues of origin

Abstract Cancer genomes are shaped by mutational processes with complex spatial variation at multiple scales. However, the underlying mechanisms of this mutagenesis and its functional and genetic determinants remain poorly understood. Somatic regional mutagenesis is known to correlate with DNA replication timing and chromatin accessibility; however, these studies have used epigenetic information from common cell lines while the epigenomes of primary human cancers remain uncharacterized in this context. Here we model megabase-scale mutation frequencies of single nucleotide variants (SNVs) in thousands of whole cancer genomes using hundreds of genome-wide profiles of chromatin accessibility and replication timing spanning primary cancer samples, normal tissues, and cell lines. Using a machine learning framework, we show that CA profiles of primary cancers, rather than those of normal cells, predict regional mutagenesis and single base substitution (SBS) signatures in most cancer types. Thus the majority of passenger mutations follow the epigenetic landscapes of transformed cells of matching cancer sites and tissues. Mutational signatures associated with carcinogen exposure, as well as signatures of unknown origin, show the strongest associations with epigenomes while associations with endogenous signatures are weaker. In most cases, overall mutation burden and SBS signatures inversely correlate with chromatin accessibility, however certain signatures are instead enriched in chromatin-accessible elements. Lastly, the genomic regions with excess mutations unexplained by the epigenetic profiles in our models likely represent a combination of localized mutagenesis and positive selection of functional oncogenic mutations. Our computational models highlight an enrichment of known driver mutations, point out novel intergenic regions with putative non-coding drivers, and indicate an over-representation of developmental and lineage-specific genes in these frequently mutated genomic regions. These results characterize the complex interplay of mutational processes, genome function, and somatic evolution in cancer and its tissues of origin, and also outline strategies for distinguishing driver and passenger mutations. Citation Format: Jüri Reimand, Oliver Ocsenas. Chromatin accessibility of primary human cancers ties regional mutational processes with tissues of origin [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1508.

Abstract 904: Neural stem cells of the subventricular zone evolve into tumors around the resection cavity after surgical resection of glioblastoma via the CXCL12/CXCR4 axis

Abstract Glioblastoma multiforme (GBM) is the most common lethal primary brain tumor. Neural stem cells (NSCs) harboring cancer-driving mutations in the subventricular zone (SVZ) have recently been reported as the cells of origin for human glioblastoma (GBM) through migration. However, the pathological role of residual tumor-initiating NSCs (tNSCs) in the SVZ is undetermined for local recurrent tumor after surgical removal of the primary tumor. After the primary resection of GBM, most recurrences still appear as local tumors around the resection cavity (RC). Accordingly, we hypothesized that, after complete removal of the primary tumor, the tNSCs in SVZ might be the source for tumor construction at RC. Here, we examined the role of tNSCs in the development of local tumors around RC using a mouse model and genetic analysis of human tissues. First, cancer-driving mutations, including Trp53, Pten, and Egfr mutations, were introduced to NSCs in SVZ through in vivo electroporation, and the RC was generated in the remote cortex through stereotaxic resection. We found that the NSCs harboring driver mutations migrated specifically to the RC through the aberrant growth of oligodendrocyte-precursor cells (OPCs). Furthermore, the SVZ-originated cells constituted the GBM around the RC at four weeks post-resection in 64% of mouse models. Second, we performed deep sequencing of longitudinal matched tissues consisting of (i) normal SVZ away from the tumor, (ii) primary tumor, (iii) recurrent tumor, and (iv) blood in the patients with GBM, IDH wildtype. We found that the dominant clones of recurrent tumors genetically arose from the SVZ, not directly from the primary tumor. Third, the CXCR4/CXCL12 axis was upregulated through migration from the SVZ and tumor formation at the RC compared with primary tumors without the RC in mouse models. Blockage of the CXCR4/CXCL12 axis decreased the number of immigrating OPC lineage cells specifically and improved tumor control and survival rates. Taken together, our results reveal that the residual tNCSs harboring cancer-driving mutations in the SVZ can re-construct tumors after surgical resection of GBM. Blocking CXCL12/CXCR4 might be a potential treatment strategy for preventing the main local failures after resection. Citation Format: Xue Li, Hyun Jung Kim, Jeong Ho Lee, Hong-Gyun Wu, Joo Ho Lee. Neural stem cells of the subventricular zone evolve into tumors around the resection cavity after surgical resection of glioblastoma via the CXCL12/CXCR4 axis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 904.

Detection and Significance of Cell-Free DNA Mutation in Pleural Effusion in Patients with Advanced NSCLC.

Objective To detect EGFR/KRAS genes in pleural effusion cell-free DNA in patients with advanced non-small-cell lung cancer (NSCLC) and to explore the clinical significance of EGFR/KRAS mutation status in pleural effusion. Methods A retrospective collection was performed on the specimens of pleural effusion and matched tissues from 50 patients with advanced NSCLC admitted to the hospital between January 2019 and January 2021. DNA mutation status of EGFR/KRAS in different specimens was detected and compared by pyrosequencing. The clinicopathological data and follow-up data of survival were collected. The relationship between DNA mutation and clinicopathological characteristics and prognosis was analyzed. Results In the 50 pleural effusion specimens, there were 22 cases (44.00%) with EGFR mutations (19/21 exon mutations), including 12 cases with EGFR19 deletion mutation and 10 cases with EGFR21 exon L858R mutation. There were 6 cases (12.00%) with KRAS mutations (single-base substitution mutations), including 4 cases with 12-codon mutation and 2 cases with 13-codon mutation. In the 50 tissue specimens, there were 24 cases (48.00%) with EGFR mutations and 4 cases (8.00%) with KRAS mutations. There was no significant difference between pleural effusion specimens and tissue specimens, with good consistency (kappa = 0.920–0.779, P > 0.05). EGFR mutation in pleural effusion was related to smoking history, types of pathological tissues, and lymph node metastasis (P < 0.05). The incidence of EGFR mutation was higher in nonsmokers, patients with lung adenocarcinoma, and patients with lymph node metastasis. The carcinoembryonic antigen (CEA) in patients with EGFR mutation was higher than that with wild-type EGFR, while the level of cytokeratin 19 fragment (Cy21-1) was lower than that with wild-type EGFR (P < 0.05). The 1-year overall survival rate in the EGFR mutation group was significantly higher than that in the EGFR wild group (68.18% vs. 42.86%) (HR = 0.419, 95% CI = 0.178–0.989, and P < 0.001). Conclusion For the detection of EGFR gene mutation, the results of the pleural effusion specimens and the tumor pathological tissue specimens were well consistent and the detection of pleural effusion could be used as an alternative method when tissue specimens cannot be obtained. EGFR gene mutations are present in majority in patients with advanced NSCLC. The incidence of EGFR mutation is higher in nonsmokers, patients with lung adenocarcinoma, those with lymph node metastasis, those with high-expression CEA, and those with low-expression Cy21-1. The prognosis is better in patients with EGFR mutation.

NCAPH is a prognostic biomarker and associated with immune infiltrates in lung adenocarcinoma

Non-SMC condensin I complex subunit H (NCAPH) plays a regulatory role in various cancers. However, its role in prognosis and immune infiltrates in lung adenocarcinoma (LUAD) remains unclear. This study examined the expression of NCAPH in tumor tissues and its association with immune infiltrates and prognostic roles in LUAD patients. Patients characteristics were obtained from The Cancer Genome Atlas (TCGA). Integrated analysis of TCGA showed that NCAPH was overexpressed across cancers, including LUAD. NCAPH expression was verified by quantitative polymerase chain reaction and western blotting in 20 LUAD matched tissues. High NCAPH expression was significantly related to T, N, M, pathologic stage, primary therapy outcome and smoking status according to the Wilcoxon rank sum test. Cox and Kaplan–Meier analyses showed that the NCAPH-high group was associated with shorter OS. The PFI and DSS in the NCAPH-high group were significantly decreased. Multivariate analysis showed that NCAPH was an independent predictive factor for poor prognosis. Gene set enrichment analysis demonstrated that the G2/M checkpoint, ncRNA metabolic, memory B cells, KRAS, E2F targets and MIER1 process were significantly associated with NCAPH expression. Single-sample Gene Set Enrichment Analysis indicated that NCAPH expression was associated with levels of Th2 and mast cells. The impact of NCAPH on malignant phenotypes was evaluated by MTT, transwell, cell cycle and apoptosis assays in vitro. The malignant phenotype of LUAD cells was inhibited if NCAPH was knocked down. In conclusion, this research indicates that NCAPH could be a potential factor for predicting prognosis and a new biomarker in LUAD.

TP53 pathogenic variants with low allele fraction in germline genetic testing.

10600 Background: Blood or saliva DNA generally considered to be representative of germline genome in genetic cancer risk assessment. However white blood cells from these samples may also include somatic origin DNA due to postzygotic variation or, most commonly, clonal hematopoiesis (CH). Low variant allele fraction (VAF) found in germline genetic testing suggest the possibility of somatic variant and may lead to misinterpretation of genetic risk. TP53, of which germline pathogenic variants are associated with Li-Fraumeni syndrome (LFS), is frequently mutated in CH. This analysis investigated characteristics of TP53 pathogenic variants with low VAF. Methods: We reviewed the prevalence and distribution of TP53 pathogenic variants(PVs) detected in 11,277 advanced cancer patients who underwent clinical testing with a clinical NGS pan-cancer panel. Potential somatic PVs were defined as variants with low VAF from 10% to 35%. The VAF were evaluated in matched tumor tissue samples if available. Results: TP53 pathogenic variants were detected in 36 (0.32%) patients from blood or saliva samples, VAF between 10% and 35% were identified in 8(22.22%) patients and 7 of them were performed NGS sequencing in matched tumor tissue samples. The average VAF of tissue samples were 9.31times lower than blood or saliva samples(21.69% vs 3.88%). Conclusions: TP53 pathogenic variants with low allele fraction in blood or saliva samples indicate the possibility of somatic variant, a reduced VAF in matched tumor tissue samples may contribute to confirmation for suspicion of somatic origin.

Leveraging personalized circulating tumor DNA (ctDNA) for detection and monitoring of molecular residual disease in high-risk melanoma.

9579 Background: High-risk melanoma has variable prognosis. Adjuvant immuno- (IO) and targeted therapy (TT) are approved for stage III-IV resected disease. However, a significant proportion of patients (pts) are cured by local treatment alone or relapse despite adjuvant therapy. Liquid biopsy with ctDNA assays have been used to predict response to treatment and identify pts at higher risk of progression/death. Personalized ctDNA assays are a highly sensitive approach that may enhance upfront risk stratification and early detection of relapse. Methods: Serial ctDNA Monitoring as a predictive Biomarker in advanced neoplAsms (SAMBA) is a Princess Margaret prospective ctDNA kinetics study (NCT03702309) in high-risk melanoma pts. Plasma is collected pre-op (pre-local treatment, if feasible), post-op (after surgery), and every 3-6 months (m) until radiological progressive disease (rPD). Personalized amplicon based NGS assays by Inivata (RaDaR) were used to detect somatic variants in ctDNA identified through whole-exome sequencing of matched tumor tissue. Progression free survival (PFS) and overall survival (OS) from the time of surgery were estimated with the Kaplan Meier and compared with the log-rank test. Results: As of December 2021, 82 of 100 planned pts have been enrolled. A total of 191 samples from 47 pts have been analyzed. Median age was 66 years (27-87), 33 were male (70%). Seven (15%), 30 (64%) and 10 (21%) were stage II/III/IV respectively. All pts had surgery and 8 (17%) adjuvant radiation. No systemic therapy was given to 11 pts (23%); 30 (64%) had IO and 6 (13%) TT. rPD occurred in 13 pts (28%). Median follow up was 24 months. A median of 48 variants were included in the personalized ctDNA panel design (35-52). ctDNA was detected (ctDNA+) at any time point in 12/47 pts (26%), of which 5/12 (42%) were BRAF and NRAS wt on tissue. Median PFS was 4.9 months (m) for ctDNA+ pts and not reached (NR) for ctDNA- pts at post-op (HR = 2.71 CI 0.60-12.31, p = 0.179). Median OS was 23.1 m vs NR in ctDNA+ vs ctDNA- pts (HR = 8.9, CI 1.45-54.77, p = 0.004). Two ctDNA+ pts had neoadjuvant IO and became ctDNA- before surgery. One, free of disease after 12 m, had ctDNA- in 4 follow up samples. The other pt was ctDNA+ in the post-op sample and relapsed within 3 m. Four of 45 (9%) pts had ctDNA+ at post-op. Two of them, including a pt who had neoadjuvant IO, did not receive adjuvant therapy and had rPD within 3 m. The other 2 pts received adjuvant IO; ctDNA cleared and pts remain free of disease at 12 and 34 m. Three pts with rising ctDNA over time experienced rPD after a median of 4 m (2-7). Conclusions: Personalized ctDNA analysis with RaDaR may improve risk of death stratification and selection of pts who could benefit from adjuvant treatment. Detection of ctDNA may precede rPD. Follow-up will continue in pts with rising ctDNA who have not yet had rPD. Pts accrual and sample collection are ongoing, and additional data will be presented. Clinical trial information: NCT03702309.

Circulating tumor DNA is a potential prognostic risk factor of recurrence in patients with hepatocellular carcinoma treated by liver transplantation.

e16196 Background: Liver transplantation (LT) is one of the most effective therapeutic strategies for hepatocellular carcinoma (HCC), however, tumour recurrence after LT often leads to poor outcomes. At present, clinical characteristic profile, imaging diagnosis as well as level of serum biomarkers prior to LT could predict the prognosis in HCC patients treated by LT. However, the sensitivity and specificity of these biomarkers are limited, highlighting the crucial need to identify novel biomarkers. Previous clinical studies have shown that postoperative circulating tumor DNA (ctDNA) can be used as a predictive biomarker for the risk of recurrence in stage I-III solid tumor. This study aimed at investigating the relationship between the level of ctDNA and the probability of recurrence in HCC patients treated by LT. Methods: HCC patients receiving LT treatment were enrolled. Peripheral blood samples before LT and available matched tissue samples were collected. Targeted sequencing of 328 cancer-related genes were conducted. The clinicopathologic data of the patients was retrospectively analyzed. The predictive efficiency for recurrence after LT of different clinicopathologic parameters and ctDNA were accessed. Results: A total of 45 patients were enrolled, including 42 males (93%) and 3 females (7%), with a median age of 51 (range 31-76 years). Thirty-five patients (78%) had detectable ctDNA before LT while gene alterations were identified in tumor tissue of all the patients. Larger tumor size (&gt; 5cm), cirrhosis, portal vein tumor thrombus (PVTT) and elevated AFP level (≥400ng/ml) prior to LT were associated with higher percentage of positive ctDNA (94% vs 73%, 84% vs 75%, 88% vs 79%, 88% vs 76%, respectively). Patients with micro-vessel invasion (MVI) had higher ctDNA abundance (maxVAF, P &lt; 0.05). Interestingly, after over 3 years’ follow-up, clinical factors including MVI, PVTT, BCLC stage, maximum tumor diameter and serum AFP failed to predict disease free survival (DFS) and overall survival (OS). In contrast, the recurrence rate was remarkably augmented (48.6% vs 0%) and the median DFS was significantly decreased (390 days vs not reached, p&lt; 0.05) in patients with detectable ctDNA. Conclusions: In conclusion, ctDNA could be a potential biomarker to sensitively predict the recurrence in patients with HCC treated by LT, which may help to improve the post-transplant management of these patients.

Identification of Novel circRNA-Based ceRNA Network Involved in the Pathogenesis of Gastric Cancer.

Objective Evidence increasingly shows that circular RNAs (circRNAs) are closely associated with tumorigenesis and cancer progression. However, the roles of circRNAs and the underlying mechanism behind these circRNAs in gastric cancer (GC) remain to be elucidated. This study is aimed at conferring a better understanding of GC pathogenesis with a specific focus on circRNA-based ceRNA action. Methods circRNA expression profiles were downloaded from two Gene Expression Omnibus (GEO) microarray datasets, GSE152309 and GSE121445. Expression profiles of miRNAs and mRNAs were collected from The Cancer Genome Atlas (TCGA) database. The ceRNA network was constructed based on circRNA-miRNA pairs and miRNA-mRNA pairs. Functional and pathway enrichment analyses were performed to evaluate functional pathways of differentially expressed mRNAs (DEmRNAs). The PPI network was constructed by mapping DEmRNAs into the STRING database to identify hub genes, and then the DEcircRNA-DEmiRNA-hub gene subnetwork was constructed. The expression levels of candidate differentially expressed circRNAs (DEcircRNAs) in cancerous and matched noncancerous gastric tissues surgically resected from 52 GC patients were determined by the RT-qPCR analysis. Results Differential expression analysis with Venn diagram analysis showed 11 overlapped DEcircRNAs (7 upregulated circRNAs and 4 downregulated ones) between cancerous tissues and noncancerous gastric tissues. The DEcircRNA-DEmiRNA-DEmRNA network was constructed, consisting of 2 DEcircRNAs, 7 DEmiRNAs, and 104 DEmRNAs. GO and KEGG pathway analyses revealed that 104 DEmRNAs might be associated with GC development and progression. The PPI network was constructed, yielding 16 hub genes, TFDP1, KRAS, LMNB1, MET, MYBL2, CDC25A, E2F5, HMGA1, HMGA2, CBFB, CBX3, CDC7, IGF2BP3, KIF11, PDGFB, and SMC1A, which were all upregulated in GC tissues compared with adjacent tumor-free gastric tissues. Based on the above hub genes in GC, the DEcircRNA-DEmiRNA-hub gene subnetwork was reconstructed based on hsa_circ_0000384 and hsa_circ_0000043, including 22 pairs of the upcircRNA-downmiRNA-upmRNA network. The expression levels of hsa_circ_0000384 and hsa_circ_0000043 were remarkably higher in GC tissues than those in matched adjacent tumor-free gastric tissues (p < 0.001), concurring with the bioinformatics results. Conclusion Our study offers a better understanding of circRNA-related ceRNA regulatory mechanism in the pathogenesis of GC, highlighting two ceRNA networks based on hsa_circ_0000384 and hsa_circ_0000043.

Sex-Differences and Associations Between Complement Activation and Synovial Vascularization in Patients with Late-Stage Knee Osteoarthritis.

PurposeSynovial inflammation in knee osteoarthritis (OA) causes disorganized synovial angiogenesis and complement activation in synovial fluid, but links between complement and synovial microvascular pathology have not been established. Since complement causes vascular pathology in other diseases and since sex-differences exist in complement activation and in OA, we investigated sex differences in synovial fluid complement factors, synovial tissue vascular pathology, and associations between complement and synovial vascular pathology in patients with late-stage knee OA.MethodsPatients with symptomatic, late-stage radiographic knee OA undergoing total knee arthroplasty or high tibial osteotomy provided matched synovial fluid and tissue biopsies during surgery. Complement factors (C2, C5, adipsin, MBL, and CFI) and terminal complement complex (sC5b-C9) were measured in synovial fluid by multiplex or enzyme-linked immunosorbent assay, respectively. Features of synovial vascular pathology (vascularization, perivascular edema, and vasculopathy) were assessed by histopathology. Multivariate linear regression models were used to assess associations between synovial fluid complement factors and histopathological features of vascular pathology, with adjustment for age, sex, body mass index, and sex interaction. Sex-disaggregated comparisons were completed.ResultsSynovial fluid biomarker and histopathology data were included from 97 patients. Most synovial fluid complement factors and synovial tissue histopathological features were similar between sexes. Synovial fluid C5 trended to lower levels in males (-20.93 ng/mL [95%CI -42.08, 0.23] p=0.05). Median vasculopathy scores (0.42 [95%CI 0.07, 0.77] p=0.02) were higher in males. In the full cohort, C5 concentration was associated with lower vascularization scores (-0.005 [95%CI -0.010, -0.0001] p=0.04) while accounting for sex*C5 interaction. In sex-disaggregated analyses, increased C5 concentration was associated with lower vascularization scores (-0.005 [95%CI –0.009, -0.0001] p=0.04) in male patients, but not in female patients. Males had higher sC5b-C9 compared to females. Additionally, males with high C5 had a higher synovial fluid concentration of sC5b-C9 compared to males with low C5. No differences were found in females.ConclusionHigher synovial fluid C5 levels were associated with increased complement activation and decreased synovial vascularization in males but not in females with OA. Future studies should test whether synovial fluid complement activation suppresses synovial angiogenesis and identify mechanisms accounting for C5-related sex-differences in synovial fluid complement activation in patients with knee OA.

Epigenetic Age Acceleration as a Risk Predictor of Hepatocellular Carcinoma

IntroductionMortality rates from hepatocellular carcinoma (HCC) have been steadily increasing in the US over the past 15 years. Most patients are diagnosed at advanced stages when the mortality rate is as high as 89%. Early diagnostic indicators are not currently available, and unfortunately, invasive liver core biopsies or resections are the standard of care to confirm HCC diagnosis. While no mutations have have proven to be tumor drivers, novel genome wide methylation studies show clear, cancer‐specific patterns of methylation and identified deregulation of epigenetic signatures. Epigenetic clocks have been shown to be a versatile tool for the prediction of a variety of patient phenotypes. Most commonly used for the prediction of age and mortality risk, these have also been used as indicators of alcohol consumption, smoking, and proxies for various blood‐based biomarkers. Acceleration in a number of clocks has been correlated to increased all‐cause mortality, as well as cancer risk. While much focus in HCC has been on clocks trained to precisely predict chronological age, confidence in the characterization of the notable epigenetic landscape and target region identification will be improved through the use of clocks trained to instead predict biological age with low signal to noise ratio. The goal of this study is to use epigenetic age acceleration to develop early cancer risk detection, improve treatment options, and to understand the relationship between accelerated liver tissue aging and tumorigenesis. Epigenetic aging events are potentially reversible adjuvant. Methylated DNA is also far more stable in serum and tissue than un‐methylated nucleic acids. These factors make epigenetic events attractive candidates for serum and biopsy biomarkers.MethodsGenomic DNA was extracted from IRB‐ approved, post‐operative or snap frozen, clinical patient liver resections or explants. DNA methylation (DNAm) was collected using Illumina BeadChip arrays. DNAm was used to calculate epigenetic age according to PCPhenoAge in over 600 liver samples, 8 tumor infiltrated NK‐cell isolates and 50 cell culture samples. PCPhenoAge is an improved version of the original biological aging clock with increased reliability and reduced susceptibility to technical noise. Multiple regression was used to residualize PCPhenoAge with respect to age and multiple phenotypes, such as alcohol consumption, known to affect epigenetic age. These residuals capture the epigenetic age acceleration independent of known behavioral risk factors for HCC.ResultsPreliminary analysis of 27 of our matched HCC patient tissues‐‐14 Alcohol and 13 Hepatitis C virus (HCV) associated HCC patients‐‐show higher PCPhenoAge in tumor versus non‐tumor tissues. Further, alcohol‐related HCC tumor tissue has significantly higher PCPhenoAge than HCV‐related HCC tumor tissue.