The Cancer Genome Atlas Cohort Research Articles

Genomic landscape and tumor mutational features of resected preinvasive to invasive lung adenocarcinoma

IntroductionAdenocarcinoma in situ (AIS) and minimally invasive adenocarcinoma (MIA) are considered pre-invasive forms of lung adenocarcinoma (LUAD) with a 5-year recurrence-free survival of 100%. We investigated genomic profiles in early tumorigenesis and distinguished mutational features of preinvasive to invasive adenocarcinoma (IAC) for early diagnosis.MethodsMolecular information was obtained from a 689-gene panel in the 90 early-stage LUAD Chinese patients using next-generation sequencing. Gene signatures were identified between pathology subtypes, including AIS/MIA (n=31) and IAC (n=59) in this cohort. Mutational and clinicopathological information was also obtained from the Cancer Genome Atlas (TCGA) as a comparison cohort.ResultsA higher mutation frequency of TP53, RBM10, MUC1, CSMD, MED1, LRP1B, GLI1, MAP3K, and RYR2 was observed in the IAC than in the AIS/MIA group. The AIS/MIA group showed higher mutation frequencies of ERBB2, BRAF, GRIN2A, and RB1. Comparable mutation rates for mutually exclusive genes (EGFR and KRAS) across cohorts highlight the critical transition to invasive LUAD. Compared with the TCGA cohort, EGFR, KRAS, TP53, and RBM10 were frequently mutated in both cohorts. Despite limited gene mutation overlap between cohorts, we observed variant mutation types in invasive LUAD. Additionally, the tumor mutation burden (TMB) values were significantly lower in the AIS/MIA group than in the IAC group in both the Chinese cohort (P=0.0053) and TCGA cohort (P<0.01).ConclusionThese findings highlight the importance of distinguishing preinvasive from invasive LUAD in the early stages of LUAD and both pathology and molecular features in clinical practice, revealing genomic tumor heterogeneity and population differences.

Identification of NINJ1 as a novel prognostic predictor for retroperitoneal liposarcoma

BackgroundRetroperitoneal liposarcoma (RPLS) is known for its propensity for local recurrence and short survival time. We aimed to identify a credible and specific prognostic biomarker for RPLS.MethodsCases from The Cancer Genome Atlas (TCGA) sarcoma dataset were included as the training group. Co-expression modules were constructed using weighted gene co-expression network analysis (WGCNA) to explore associations between modules and survival. Survival analysis of hub genes was performed using the Kaplan–Meier method. In addition, independent external validation was performed on a cohort of 135 Chinese RPLS patients from the REtroperitoneal SArcoma Registry (RESAR) study (NCT03838718).ResultsA total of 19 co-expression modules were constructed based on the expression levels of 26,497 RNAs in the TCGA cohort. Among these modules, the green module exhibited a positive correlation with overall survival (OS, p = 0.10) and disease-free survival (DFS, p = 0.06). Gene set enrichment analysis showed that the green module was associated with endocytosis and soft-tissue sarcomas. Survival analysis demonstrated that NINJ1, a hub gene within the green module, was positively associated with OS (p = 0.019) in the TCGA cohort. Moreover, in the validation cohort, patients with higher NINJ1 expression levels displayed a higher probability of survival for both OS (p = 0.023) and DFS (p = 0.012). Multivariable Cox analysis further confirmed the independent prognostic significance of NINJ1.ConclusionsWe here provide a foundation for the establishment of a consensus prognostic biomarker for RPLS, which should not only facilitate medical treatment but also guide the development of novel targeted drugs.

Molecular alterations in claudin 18 suppressed and non-suppressed gastric adenocarcinomas to guide targeted therapies

ABSTRACT Background Gastric adenocarcinoma represents an aggressive type of cancer and an important cause of cancer mortality. Progress in gastric cancer therapeutics has resulted from a better understanding of the molecular pathogenesis of the disease and introduction of targeted therapies, but most gastric cancer patients still rely on non-targeted chemotherapy as the mainstay of treatment for advanced disease. Methods An analysis of publicly available series from The Cancer Genome Atlas (TCGA) gastric cancer cohort was undertaken to delineate the clinical and genomic landscape of gastric cancers with suppressed expression of claudin 18 compared with cancers with non-suppressed claudin 18. Claudin 18 suppressed cancers were defined as having an mRNA expression z-score relative to normal samples (log RNA Seq V2) of less than −1. Claudin 18 non-suppressed cancers were defined as having an mRNA expression z-score relative to normal samples (log RNA Seq V2) above 0.5. Results Gastric cancers with claudin 18 mRNA suppression represented 7.7% of the gastric adenocarcinomas of TCGA cohort, while non-suppressed cancers represented 46.6% of the cases. The two groups did not differ in clinical and genomic characteristics, such as mean age, histology, grade, and stage. The mutation landscape of claudin 18 suppressed cases included high mutation rates of TP53, of genes of the WNT/β-catenin pathway and of ubiquitin ligase FBXW7. Moreover, a subset of both claudin 18 suppressed and non-suppressed cancers displayed mutations in Mismatch Repair (MMR) associated genes or a high tumor mutation burden (TMB). At the mRNA expression level, claudin 18 suppressed gastric cancers showed up-regulation of EMT core transcription factor Snail 2 and down-regulation of genes of HLA cluster. The survival of gastric cancer patients with claudin 18 mRNA suppression was not significantly different compared with patients with non-suppressed claudin 18. Conclusion Sub-sets of gastric cancers with claudin 18 mRNA suppression displayed characteristics of potential therapeutic interest, such as mutations in WNT and PI3K pathways and MMR defects. These may guide the development of alternative targeted therapies, in this sub-set of gastric cancers which are not candidates for claudin 18 targeting therapies.

Characterization of zinc finger protein 536, a neuroendocrine regulator, using pan-cancer analysis

BackgroundPrevious studies suggested that zinc finger protein 536 (ZNF536) was abundant in the central brain and regulated neuronal differentiation. However, the role of ZNF536 in cancer has remained unclear.MethodsZNF536 mutation, copy number alteration, DNA methylation, and RNA expression were explored using public portals. Data from The Cancer Genome Atlas (TCGA) were utilized to analyze pathways and tumor microenvironment (TME), with a focus on prognosis in both TCGA and immunotherapy pan-cancer cohorts. Methylated ZNF536 from small cell lung cancer (SCLC) cell lines were utilized to train with probes for conducting enrichment analysis. Single-cell RNA profile demonstrated the sublocalization and co-expression of ZNF536, and validated its targets by qPCR.ResultsGenetic alterations in ZNF536 were found to be high-frequency and a single sample could harbor different variations. ZNF536 at chromosome 19q12 exerted a bypass effect on CCNE1, supported by CRISPR data. For lung cancer, ZNF536 mutation was associated with longer survival in primary lung adenocarcinoma (LUAD), but its prognosis was poor in metastatic LUAD and SCLC. Importantly, ZNF536 mutation and amplification had opposite prognoses in Stand Up To Cancer-Mark Foundation (SU2C-MARK) LUAD cohort. ZNF536 mutation altered the patterns of genomic alterations in tumors, and had distinct impacts on the signaling pathways and TME compared to ZNF536 amplification. Additionally, ZNF536 expression was predominantly in endocrine tumors and brain tissues. High-dimensional analysis supported this finding and further revealed regulators of ZNF536. Considering that the methylation of ZNF536 was involved in the synaptic pathway associated with neuroendocrine neoplasms, demonstrating both diagnostic and prognostic value. Moreover, we experimentally verified ZNF536 upregulated neuroendocrine markers.ConclusionsOur results showed that ZNF536 alterations in cancer, including variations in copy number, mutation, and methylation. We proved the involvement of ZNF536 in neuroendocrine regulation, and identified highly altered ZNF536 as a potential biomarker for immunotherapy.

Abstract PO3-25-03: Can intratumoral microRNA-125a levels be a prognostic biomarker of a ER-positive breast cancer patients？

Abstract Background: MicroRNA-125a (miR-125a) has been shown to function as tumor suppressive miRNA by pre-clinical in vitro studies. However, to the best of our knowledge, there has been no study that investigated the clinical relevance of miR-125a in breast cancer patients. We hypothesized that miR-125a expression in breast cancer associate with its aggressiveness and can be a prognostic biomarker. Material and Methods: Non-tumorigenic epithelial cell line, MCF-10A and breast cancer cell lines, MCF-7, T47D, and MDA-MB231, were used to analyze the expression levels of miR-125a as well as cell viability. The data of 2042 patients were extracted from publicly available databases, The Molecular Taxonomy of Breast Cancer International Consortium (METARBRIC), The Cancer Genome Atlas (TCGA), and GSE57897. These databases included clinicopathological and survival data as well as transcriptome data. The survival analysis, gene set enrichment analysis (GSEA) were conducted comparing miR-125a high expressing and low expressing tumors, divided by the median cutoff. The association between the miR-125a expression and tumor immune microenvironment was assessed by xCell. Results: The expression levels of miR-125a were suppressed in breast cancer cell lines, MCF-7, T47D, and MDA-MB231compared with MCF-10A. This was concordant with human samples. The expression levels of miR-125a were lower in tumors compared with normal breast tissues in both TCGA and GSE57897 cohorts. These results were in agreement with the notion that it is a tumor suppressive miRNA (p< 0.001 and p< 0.001, respectively). ER-positive/HER2-negative (ER+/HER2-) showed the highest miR-125a expression among the subtypes in TCGA (p< 0.001). MiR-125a expression was not associated with cancer staging in any of the subtypes in neither TCGA nor METABRIC cohorts. To our surprise, high miR-125a tumors demonstrated worse disease free survival (DFS), disease specific survival (DSS), and overall survival (OS) compared with low expressing in ER+/HER2- breast cancer patients (p=0.008, p=0.005, and p=0.037) which was not the case for the other subtypes in METABRIC cohort. Interestingly, miR-125a expression was associated with Nottingham histological grade only in ER+/HER2- among the subtypes (p< 0.001). Also, high miR-125a tumors significantly demonstrated higher expression of MKI67, one of the most commonly used parameters for cell proliferation, correlated with miR-125a expression in ER+/HER2- (p=0.02), but not in other subtypes. Furthermore, high miR-125a tumors enriched four out of five cell proliferation-related gene sets in Hallmark collection, such as E2F Targets, G2M Checkpoint, Mitotic Spindle, and MYC Targets V2 in ER+/HER2- subtype, but not in TNBC. This was also the case in immune-related gene sets; interferon-alpha response and interferon-gamma response that enriched to high miR-125a tumors in ER+/HER2-, but not in TNBC. Anti-tumor immune cells, CD8 cells, T-helper type 1 cells, and M1 macrophages, were all elevated in high miR-125a tumors in ER+/HER2- subtype (all p< 0.03), but none in TNBC. Also, pro-cancerous immune cell such as T-helper type 2 cells were similarly highly infiltrated with high miR-125a tumors only in ER+/HER2- subtype (p=0.026). Given its association with the survival outcome, we cannot help but speculate that miR-125a expression is parallel with highly proliferative ER+/HER2- breast cancer, but its tumor suppressive effect is not enough to improve survival outcome. Conclusion: High miR-125a levels in tumors were associated with aggressive biology and worse survival, thus it can be a candidate for a prognostic biomarker in ER+/HER2- patients. Citation Format: Yoshihisa Tokumaru, Masanori Oshi, Mai Okawa, Saki Ando, Yoshimi Niwa, Ryutaro Mori, Kazuaki Takabe, Manabu Futamura. Can intratumoral microRNA-125a levels be a prognostic biomarker of a ER-positive breast cancer patients？ [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO3-25-03.

Abstract PO2-08-05: ERBB2 mutated breast cancer detected by next generation sequencing: integrated analysis based on real-world data and The Cancer Genome Atlas

Abstract Background ERBB2 mutation has been recognized as an actionable target for pan-HER tyrosine kinase inhibitors in the era of precision medicine for solid tumors. In our real-world cohort utilizing next-generation sequencing (NGS), we successfully identified cases of ERBB2-mutated breast cancer. Furthermore, we investigated the clinical and pathological characteristics of these cases by integrating our data with The Cancer Genome Atlas (TCGA). Methods Our real-world cohort with NGS included 423 breast cancer patients treated at Gangnam Severance Hospital in Seoul, Republic of Korea, between April 2017 and January 2023. Among these patients, 13 were confirmed to have ERBB2 mutation through NGS analysis. We also incorporated data from TCGA database, which included 818 cases with ERBB2 sequencing. We compared the clinical characteristics (age, sex, stage, and ER, PR, and HER2 status) between the two groups. Furthermore, we examined the frequency of two of the most common mutations, PIK3CA and TP53, in both groups. Additionally, we investigated the distribution of PAM50 subtypes in the TCGA cohort. Results This study analyzed a total of 1,244 cases with NGS data. Among these cases, 31 (2.5%) exhibited ERBB2 mutation, with 13 patients identified in our dataset and 18 in the TCGA data. Among the 31 cases with ERBB2 mutation, 21 (67.7%) had HER2-negative disease. No significant differences were observed in other clinical characteristics based on ERBB2 status. Within the group of 31 ERBB2-mutated cases, TP53 mutations were found in 12 cases (38.7%), while PIK3CA mutations were found in 12 cases (38.7%). In the TCGA dataset, 4 cases (22.2%) with ERBB2 mutation were classified as HER2-E subtype, and the rate of HER2-E subtype was significantly higher in the ERBB2-mutated group compared to the ERBB2-wild group (22.2% vs. 7.6%, p=0.024). Conversely, 14 cases (77.8%) with ERBB2 mutation exhibited a non-HER2 enriched subtype. Conclusions Consistent with previous knowledge, ERBB2 mutations can be detected in HER2-negative breast cancer, despite their rarity. Furthermore, a considerable proportion of ERBB2-mutated breast cancer may depend on various oncogenic signaling pathways other than the HER2 pathway. Further studies are warranted to optimize therapeutic strategies for ERBB2-mutated breast cancer. Citation Format: Junghyun Kim, Yoonwon Kook, Seung Ho Baek, Min Ji Kim, Sohyun Moon, Seung Eun Lee, Jee Hung Kim, Soong June Bae, Joon Jeong, Sung Gwe Ahn. ERBB2 mutated breast cancer detected by next generation sequencing: integrated analysis based on real-world data and The Cancer Genome Atlas [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-08-05.

Identification of ATM Mutation as a Potential Prognostic Biomarker for Immune Checkpoint Inhibitors Therapy

Background: Ataxia telangiectasia mutated (ATM), an apical DNA damage response gene, is a commonly mutated gene in tumors, and its mutation could strengthen tumor immunogenicity and alter the expression of PD-L1, which potentially contributes to immune checkpoint inhibitors (ICIs) therapy. Methods: The characteristics of ATM mutation and its relationship with the ICIs-treated clinical prognosis have been analyzed comprehensively in this paper. The overall frequency of ATM mutations has been found to be 4% (554/10953) in the cancer genome atlas (TCGA) cohort. Results: Both the TMB and MSI levels in patients with ATM mutations were significantly higher than those in patients without mutations (P < 0.0001). The median TMB was positively correlated with the frequency of ATM mutations (r = 0.54, P = 0.003). In the TCGA cohort, patients with ATM mutations had better clinical benefits in terms of overall survival (OS, hazard ratio (HR) = 0.736, 95% CI = 0.623 - 0.869), progression-free survival (PFS, HR = 0.761, 95% CI = 0.652 - 0.889), and disease-free survival (DFS, HR = 0.686, 95% CI = 0.512 - 0.919)] than patients without ATM mutations. Subsequently, the verification results showed ATM mutations to be significantly correlated with longer OS in ICIs-treated patients (HR = 0.710, 95% CI = 0.544 - 0.928). Further exploration indicated ATM mutation to be significantly associated with regulated anti-tumor immunity (P < 0.05). Conclusion: Our findings highlight the value of ATM mutation as a promising biomarker to predict ICIs therapy in multiple tumors.

Increased co-expression of ICOS and PD-1 predicts poor overall survival in patients with acute myeloid leukemia

BackgroundInducible co-stimulatory factor (ICOS) has a dual role: activating cytotoxic T cells against tumors or exacerbating immunosuppression of regulatory T cells (Tregs) to participate in immune evasion. However, the correlation between ICOS and its co-expression with inhibitory immune checkpoints (IICs) and prognosis in acute myeloid leukemia (AML) is little known. MethodsThe prognostic importance of ICOS and IICs in 62 bone marrow (BM) samples of de novo AML patients from our clinical center (GZFPH) was explored and then the RNA sequencing data of 155 AML patients from the Cancer Genome Atlas (TCGA) database was used for validation. ResultsIn both GZFPH and TCGA cohorts, high expression of ICOS was significantly associated with poor overall survival (OS) in patients with AML (P < 0.05). Importantly, co-expression of ICOS and PD-1, PD-L1, PD-L2, CTLA-4, and LAG-3 predicted poor OS in AML; among them, ICOS/PD-1 was the optimal combination of immune checkpoints (ICs). The co-expression of ICOS and PD-1 was correlated with poor OS in non-acute promyelocytic leukemia (non-APL) patients following chemotherapy. Additionally, ICOS/PD-1 was an independent OS-predicting factor (P < 0.05). Notably, a nomogram model was constructed by combining ICOS/PD-1, age, European Leukemia Net (ELN) risk stratification, and therapy to visually and personalized predict the 1-, 3-, and 5-year OS of patients with non-APL. ConclusionIncreased expression of ICOS predicted poor outcomes, and ICOS/PD-1 was the optimal combination of ICs to predict outcomes in patients with AML, which might be a potential immune biomarker for designing novel AML therapy.

A radiogenomic multimodal and whole-transcriptome sequencing for preoperative prediction of axillary lymph node metastasis and drug therapeutic response in breast cancer: a retrospective, machine learning and international multicohort study.

Axillary lymph nodes (ALN) status serves as a crucial prognostic indicator in breast cancer (BC). The aim of this study was to construct a radiogenomic multimodal model, based on machine learning and whole-transcriptome sequencing (WTS), to accurately evaluate the risk of ALN metastasis (ALNM), drug therapeutic response and avoid unnecessary axillary surgery in BC patients. In this study, conducted a retrospective analysis of 1078 BC patients from The Cancer Genome Atlas (TCGA), The Cancer Imaging Archive (TCIA), and Foshan cohort. These patients were divided into the TCIA cohort ( N =103), TCIA validation cohort ( N =51), Duke cohort ( N =138), Foshan cohort ( N =106), and TCGA cohort ( N =680). Radiological features were extracted from BC radiological images and differentially expressed gene expression was calibrated using technology. A support vector machine model was employed to screen radiological and genetic features, and a multimodal model was established based on radiogenomic and clinical pathological features to predict ALNM. The accuracy of the model predictions was assessed using the area under the curve (AUC) and the clinical benefit was measured using decision curve analysis. Risk stratification analysis of BC patients was performed by gene set enrichment analysis, differential comparison of immune checkpoint gene expression, and drug sensitivity testing. For the prediction of ALNM, rad-score was able to significantly differentiate between ALN- and ALN+ patients in both the Duke and Foshan cohorts ( P <0.05). Similarly, the gene-score was able to significantly differentiate between ALN- and ALN+ patients in the TCGA cohort ( P <0.05). The radiogenomic multimodal nomogram demonstrated satisfactory performance in the TCIA cohort (AUC 0.82, 95% CI: 0.74-0.91) and the TCIA validation cohort (AUC 0.77, 95% CI: 0.63-0.91). In the risk sub-stratification analysis, there were significant differences in gene pathway enrichment between high and low-risk groups ( P <0.05). Additionally, different risk groups may exhibit varying treatment responses ( P <0.05). Overall, the radiogenomic multimodal model employs multimodal data, including radiological images, genetic, and clinicopathological typing. The radiogenomic multimodal nomogram can precisely predict ALNM and drug therapeutic response in BC patients.

High Ki67 Gene Expression Is Associated With Aggressive Phenotype in Hepatocellular Carcinoma.

Hepatocellular carcinoma (HCC) with high Ki67 protein expression, the most commonly used cell proliferation marker, is associated with an aggressive biologic phenotype; however, conventional immunostaining is hampered by variability in institutional protocol, specific antibody probe, and by assessor subjectivity. To this end, we hypothesized that Ki67 gene (MKi67) expression would identify highly proliferative HCC, and clarify its association with oncologic outcome, tumor progression, and immune cell population in the tumor microenvironment (TME). Furthermore, we sought to identify the cell-cycle gene expression profile that confers this aggressive phenotype. A total of 473 HCC patients with clinicopathological data associated with transcriptome were selected for this study: 358 patients from The Cancer Genome Atlas (TCGA) as the testing cohort, and 115 from GSE76427 as the validation cohort. Each cohort was divided into a highly proliferative group (MKi67-high) and the low MKi67 group (MKi67-low) by the median of Ki67 gene (MKi67) expression levels. MKi67-high HCC patients had worse disease-free survival (DFS), disease-specific survival (DSS), and overall survival (OS) independent of histological grade in the TCGA cohort. MKi67 expression correlated with histological grade and tumor size. MKi67 expression increased throughout the HCC carcinomatous sequence from normal liver, cirrhotic liver, early HCC, and advanced HCC. MKi67-high HCC was associated with higher intratumor heterogeneity, homologous recombination deficiency, and altered fraction as well as intratumoral infiltration of T helper type 1 (Th1) and Th2 cells, but lower interferon-gamma response and M2 macrophage infiltration. Cell proliferation-related gene sets in the Hallmark collection (E2F targets, G2M checkpoint, Myc target v1 and mitotic spindle), MTORC1 signaling, DNA repair, PI3K MTOR signaling, and unfolded protein response were all enriched in the MKi67-high HCC (false discovery rate (FDR) < 0.25). High MKi67 gene expression identified highly proliferative HCC with aggressive biology involving classical pathways in cell cycle regulation and DNA repair, as well as poor overall oncologic outcomes. This suggests potential for personalized treatment strategies, but validation and refinement of these observations require further research to elucidate the underlying mechanisms and validate therapeutic targeting of these pathways in MKi67-high HCC tumors.

Identification of Novel Stemness-based Subtypes and Construction of a Prognostic Risk Model for Patients with Lung Squamous Cell Carcinoma.

Although cancer stem cells (CSCs) contribute to tumorigenesis, progression, and drug resistance, stemness-based classification and prognostic signatures of lung squamous cell carcinoma (LUSC) remain unclarified. This study attempted to identify stemness-based subtypes and develop a prognostic risk model for LUSC. Based on RNA-seq data from The Cancer Genome Atlas (TCGA), Gene-Expression Omnibus (GEO) and Progenitor Cell Biology Consortium (PCBC), mRNA expression-based stemness index (mRNAsi) was calculated by one-class logistic regression (OCLR) algorithm. A weighted gene coexpression network (WGCNA) was employed to identify stemness subtypes. Differences in mutation, clinical characteristics, immune cell infiltration, and antitumor therapy responses were determined. We constructed a prognostic risk model, followed by validations in GEO cohort, pan-cancer and immunotherapy datasets. LUSC patients with subtype C2 had a better prognosis, manifested by higher mRNAsi, higher tumor protein 53 (TP53) and Titin (TTN) mutation frequencies, lower immune scores and decreased immune checkpoints. Patients with subtype C2 were more sensitive to Imatinib, Pyrimethamine, and Paclitaxel therapy, whereas those with subtype C1 were more sensitive to Sunitinib, Saracatinib, and Dasatinib. Moreover, we constructed stemness-based signatures using seven genes (BMI1, CCDC51, CTNS, EIF1AX, FAM43A, THBD, and TRIM68) and found high-risk patients had a poorer prognosis in the TCGA cohort. Similar results were found in the GEO cohort. We verified the good performance of risk scores in prognosis prediction and therapy responses. The stemness-based subtypes shed novel insights into the potential roles of LUSC-stemness in tumor heterogeneity, and our prognostic signatures offer a promising tool for prognosis prediction and guide therapeutic decisions in LUSC.

DNMT3A promotes glioma growth and malignancy via TNF-α/NF-κB signaling pathway.

DNMT3A is the main molecule responsible for DNA methylation in cells. DNMT3A affects the progression of inflammation, degenerative diseases, and malignant tumors, and exhibits significant aberrantly expression in tumor tissues. Transcriptome data and relevant clinical information were downloaded from The Cancer Genome Atlas (TCGA), Chinese Glioma Genome Atlas (CGGA), and Gene Expression Omnibus (GEO) datasets. Differential expression analysis and prognostic analysis were conducted based on above statistics. We constructed a clinical prognostic model and identified DNMT3A as an independent prognostic factor to accurately predict patient prognosis. Differential gene enrichment analysis revealed that DNMT3A affects the progression of glioma through multiple pathways, among which the tumor necrosis factor-α (TNF-α)/nuclear factor-kappa B (NF-κB) pathway shows a strong correlation. Immunological analysis also revealed a certain correlation between DNMT3A and tumor immunity. We demonstrated through gene editing that DNMT3A can affect the release of TNF-α in cells, thereby affecting the progression of glioma. Functional experiments have also demonstrated that DNMT3A plays a crucial role in tumors. RNA-sequencing and survival analyses of lower-grade glioma (LGG) patients in TCGA, CGGA, and GEO cohorts showed that high DNMT3A expression correlated with poor prognosis of LGG patients. Univariate and multivariate Cox regression analyses showed that DNMT3A expression was an independent prognostic indicator in LGG. The prognosis prediction nomogram with age, World Health Organization (WHO) grading, and DNMT3A expression showed reliable performance in predicting the 1-, 3-, and 5-year overall survival (OS) of LGG patients. Functional enrichment analysis, gene set enrichment analysis (GSEA), and ESTIMATE algorithm analyses showed that DNMT3A expression was associated with the tumor infiltration of immune cells and predicted response to immunotherapy in two immunotherapy cohorts of pan-cancer patients. Furthermore, short hairpin RNA (shRNA)-mediated knockdown of DNMT3A in the LGG cell lines suppressed proliferation, migration, and invasion of LGG cells by downregulating the TNF-α/NF-κB signaling pathway. Our data showed that DNMT3A was a potential prognostic biomarker in glioma. DNMT3A promoted proliferation and malignancy of LGG cells through the TNF-α/NF-κB signaling pathway. DNMT3A is a promising therapeutic target for treating patients with LGG.

Abstract 6125: Differential expression of chemokine genes by race and Breast Cancer Consensus Subtypes in hormone receptor positive breast cancer

Abstract Hormone receptor positive (HR+) breast cancer is a heterogeneous subtype and African American women (AAW) are twice as likely to have HR+ tumors with lower levels of HR staining (1-10%) compared to European American women (EAW). These tumors are associated with substantially worse survival compared to strongly HR+ tumors. We further hypothesized that HR+ tumors from AAW would be more likely to have molecular features typically associated with HR-negative breast cancer and poor prognosis. To test this, we used the Breast Cancer Consensus Subtyping (BCCS) method and evaluated expression of chemokine receptors known to be differentially expressed by race, PAM50 subtypes, and survival. We downloaded The Cancer Genome Atlas (TCGA) expression data for 394 female EA and AA HR+/HER2- breast cancer patients using the cBio Cancer Genomics Portal. Breast Cancer Consensus Subtypes (BCCS) were determined using the R program BCCSclassifier. BCCS classification results in five subtypes: ER-negative non-basal-like (BCS1), ER-negative basal-like (BCS2), ER+ highly proliferative with poor prognosis (BCS3), ER+ stromal infiltration with good prognosis (BCS4), and ER+ hormone response genes highly expressed with good prognosis (BCS5). High and low expression of chemokines (ACKR1, ACKR4, CCR3, CCR6, CCRL2, CXCR1, CXCR2, CXCR4, CXCR6, CX3CR1) was dichotomized based on median expression of each gene within our study cohort. Distribution of BCCS by race was assessed with multinomial logistic regression. Associations between chemokine gene expression, BCC subtype and race were evaluated using logistic regression. An alpha value of 0.05 was set to determine statistical significance. From the TCGA cohort, AAW were more likely to have the BCS2 ER-negative basal-like phenotype (OR=7.17, p&amp;lt;0.001) as well as the poor prognosis highly proliferative ER+ BCS3 phenotype (OR=2.83, p=0.036) versus the high hormone expression BCS5 subtype. In multivariable models evaluating both race and BCCS subtype, AAW were 6.17 times more likely to have lower than median expression of CX3CR1 (p&amp;lt;0.001) and CXCR2 (OR=2.23, p=0.0096) compared to EAW. Notably, all chemokine receptors consistently showed significantly lower expression in BCS1-4 subtypes compared to BCS5. For CCRL2, the effect estimates for BCS4 vs. BCS5 significantly differed by race (p=0.039). In AAW, BCS4 was associated with higher than median expression (OR=1.99, p=0.49), although not statistically significant. In EAW, BCS4 was associated with lower than median expression (0.23, p&amp;lt;0.001). Our findings show that AAW with HR+ breast cancer have tumors with more inherently aggressive molecular subtypes associated with poor prognosis compared to EAW. These results also further support the premise that HR+/HER2- tumors are highly heterogeneous and strongly suggest that chemokine gene expression is associated with breast cancer subtype. Citation Format: Abigail M. Fielder, Manohar Ratnam, Gregory Dyson, Kristen Purrington. Differential expression of chemokine genes by race and Breast Cancer Consensus Subtypes in hormone receptor positive breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6125.

Abstract 6548: Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples

Abstract The NCI's The Cancer Genome Atlas (TCGA) project profiled over 10,000 tumor samples over the course of 10 years. As different tissue-specific working groups reviewed all of the available data, these patient samples were separated into distinct molecular subtypes, and these clusters were reported in various marker papers. While these assignments provided invaluable information about the common patterns of molecular characteristics in different types of cancer there was no consistent methodology for assigning new samples to these defined molecular subtypes.The NCI's Tumor Molecular Pathology group was formulated to create machine learning-based models that could be applied to non-TCGA samples and determine their TCGA mapped subtypes. Five modeling systems, JADBio, SKGrid by the Oregon Health and Science University, CloudForest by the Institute of Systems Biology, AKLIMATE by University of California Santa Cruz and subSCOPE by BC Cancer’s Genome Sciences Centre, were trained to recognize TCGA subtypes using multi-omic measurements from gene expression, DNA methylation, miRNA expression, copy number, and somatic mutation calls. While the TCGA samples were profiled using multi-omic technologies, single platform and/or compact feature set models also were assessed for their ability to assign these classifications. Each machine learning system created predictive models for 106 subtypes from 26 cancer types using as few features as possible, with a maximum of 100 features allowed for scored models. A set of 411,706 models was developed, composed of results of each of the learning methods across the various omic platforms. Top models, both multi-omic and single platform, were selected for each cancer type. On average, models were able to achieve an overall weighted F1 score of 0.895 with 42 features. While the top models for each cancer type had an overall weighted F1 mean performance of 0.936 with a mean of 29 features, in 20 of the 26 cancer types models using only gene expression provided the best performance. Analysis of features selected by the models showed some known onco-drivers were selected by many models, but many times different models would utilize features of different genes with similar levels of performance. Network-level analysis revealed that many genes of these selected features operated within the same pathways.Transferability of these models to external datasets was tested, taking TCGA breast cancer trained models and applying them to AURORA and METABRIC datasets. Interestingly, despite the data platform difference between TCGA (RNAseq) and METABRIC (microarray), model performance saw only minimal degradation of F1 values in transfer. This set of models and the training dataset will provide new opportunities for researchers and translational scientists to connect new tumors to the subtypes seen in the TCGA cohorts. Citation Format: Kyle Ellrott, Chris K. Wong, Christina Yau, Mauro A. Castro, Jordan Lee, Brian Karlberg, Jasleen K. Grewal, Vincenzo Lagani, Bahar Tercan, Verena Friedl, Toshinori Hinoue, Vladislav Uzunangelov, Lindsay Westlake, Xavier Loinaz, Ina Felau, Peggy Wang, Anab Kemal, Samantha J. Caesar-Johnson, Ilya Shmulevich, Alexander J. Lazar, Ioannis Tsamardinos, Katherine A. Hoadley, The Cancer Genome Atlas Analysis Network, Gordon A. Robertson, Theo A. Knijnenburg, Christopher C. Benz, Joshua M. Stuart, Jean C. Zenklusen, Andrew D. Cherniack, Peter W. Laird. Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6548.

Abstract 5064: Wild type KRAS amplification as de novo oncogenic driver in non-small cell lung cancer

Abstract Introduction: Oncogenic KRAS mutations define subsets of non-small cell lung cancer (NSCLC) with unique clinicopathologic and genomic features, and the recent development of direct KRAS inhibitors has changed the treatment landscape of these patients (pts). However, whether KRAS amplification is oncogenic and defines unique subset of potentially targetable NSCLCs is currently unknown. Methods: Clinicopathologic and genomic data were abstracted from multiple independent cohorts of NSCLCs which underwent tumor genomic profiling at the Dana-Farber Cancer Institute (Cohort 1), AACR-GENIE v.13, TCGA, and other 212 studies (Cohort 2). Whole transcriptome sequencing and quantitative proteomics data from the The Cancer Genome Atlas (TCGA) and the Cancer Cell Lines Encyclopedia (CCLE) were used to correlate KRAS amplification with KRAS RNA levels and protein expression. KRAS doxycycline-inducible models were generated to determine in vitro and in vivo oncogenicity of wild-type KRAS overexpression. Results: Among 15,341 pts with NSCLC identified, KRAS amplification was identified in 355 (2.4%) cases. Compared to KRAS not amplified cases, pts with KRAS amplified NSCLC were more likely to be males (52.4% vs 45.6%, p=0.01), have a history of smoking (82.1% vs 64.2%, p&amp;lt;0.01), and higher median pack-years (35 vs 23.5, p&amp;lt;0.01). KRAS amplification was also associated with higher median aneuploidy levels (as fraction of genome altered) (30.4% vs 14.3%, p&amp;lt;0.01), tumor mutational burden (p&amp;lt;0.01), and increased median PD-L1 expression (20% vs 5%, p=0.01). KRAS amplified cases were enriched with KRAS, POLE2, SLC34A2 and PALB2 mutations, while EGFR mutations were mutually exclusive (Q&amp;lt;0.1). Among NSCLC samples with transcriptomic and proteomic profiling from the TCGA and CCLE cohorts, KRAS amplification was associated with significantly increased KRAS mRNA (p&amp;lt;0.01) and protein expression (p&amp;lt;0.01), compared to KRAS diploid samples. Among pts with clinical outcomes data available (N=9,335), median overall survival (OS) from the date of diagnosis was significantly shorter in pts with KRAS amplified vs KRAS diploid genotype in both KRAS mutant (HR: 0.71, P=0.02) and KRAS WT (HR: 0.74, P=0.02) NSCLCs. In KRAS doxycycline-inducible models (BEAS-2B cells) we demonstrated increased cell proliferation and MAPK activation upon doxycycline-induced KRAS expression. In vivo, sustained expression of KRAS in animals constantly fed with doxycycline-diet triggered tumor formation and development of metastasis, indicating that WT KRAS overexpression is oncogenic. Conclusion: KRAS amplification defines a novel subset of NSCLCs characterized by distinct clinicopathologic and genomic features, and worse survival. Developing novel therapeutic strategies to target KRAS amplification may improve outcomes in pts with NSCLC. Citation Format: Biagio Ricciuti, Sandra V. Michelina, Joao Alessi, Xinan Wang, Federica Pecci, Alessandro Di Federico, Malini Gandhi, Giuseppe Lamberti, Enrico Patrucco, Lynette Sholl, Andrew Aguirre, Chiara Ambrogio, Mark Magdi Awad. Wild type KRAS amplification as de novo oncogenic driver in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5064.

Association of tumour mutation burden with prognosis and its clinical significance in stage III gastric cancer

Background: To explore the correlation between the tumour mutation burden (TMB) and prognosis and its clinical significance among patients with stage III gastric cancer (GC). Methods: Patients with stage III GC were divided into a high TMB and low TMB group in both a study cohort of 38 patients and the Cancer Genome Atlas (TCGA) cohort of 173 patients. In the study cohort, next-generation sequencing was used to detect mutated GC genes and obtain TMB data. In the TCGA cohort, gene set enrichment analysis was performed, and the relationship between TMB, prognosis and clinicopathologic factors was analysed. Western blot and quantitative real-time polymerase chain reaction were used to detect the expression levels of both proteins and genes. Cell viability was measured using methyl thiazolyl tetrazolium and transwell cell assays. Results: Patients in the high TMB group had better overall survival (OS) rates than patients in the low TMB group for both cohorts and TMB was associated with age, mutation signature 1 and mutation signature 17. The Cox regression analysis revealed that age, not TMB, was an independent prognosis factor. Furthermore, genes with high-frequency mutations were significantly enriched in the Notch and RTK-RAS signalling pathways. The activation of these pathways was lower in the high TMB compared with the low TMB group, and the proliferation and migration abilities of GC cells showed a similar pattern in both TMB groups. Conclusion: Patients in the high TMB group had better OS rates than patients in the low TMB group. Genes with high-frequency mutations were significantly enriched in the RTK-RAS and Notch pathways. Hence, TMB could serve as a prognosis biomarker with potential clinical significance.

Macrophage migration inhibitory factor (MIF) predicts survival in patients with clear cell renal cell carcinoma.

Clear cell renal cell carcinoma (ccRCC) is one of the most common subtypes of renal cancer, with 30% of patients presenting with systemic disease at diagnosis. This aggressiveness is a consequence of the activation of epithelial-mesenchymal transition (EMT) caused by many different inducers or regulators, signaling cascades, epigenetic regulation, and the tumor environment. Alterations in EMT-related genes and transcription factors are associated with poor prognosis in ccRCC. EMT-related factors suppress E-cadherin expression and are associated with tumor progression, local invasion, and metastasis. The aim of this study was to investigate the expression levels and prognostic significance of macrophage migration inhibitory factor (MIF), β-catenin, and E-cadherin in ccRCC patients. We examined these proteins immunohistochemically in tumor areas and adjacent normal tissues resected from patients with ccRCC. Analysis of the cancer genome atlas (TCGA) cohort was performed to verify our results. Kaplan-Meier analysis showed that median overall survival (OS) was significantly shorter in patients with tumors exhibiting high MIFn and MIFm-c levels compared to those with low MIFn and MIFm-c levels (p = 0.03 and p = 0.007, respectively). In the TCGA cohort, there was a significant correlation between MIF expression and OS (p < 0.0001). In conclusion, this study provides further evidence for the biological and prognostic value of MIF in the context of EMT as a potential early prognostic marker for advanced-stage ccRCC.

Tumor-Stroma Proportion to Predict Chemoresistance in Patients With Ovarian Cancer

Platinum-based chemotherapy is the backbone of standard-of-care treatment for patients with advanced-stage, high-grade serous carcinoma (HGSC), the most common form of ovarian cancer; however, one-third of patients have or acquire chemoresistance toward platinum-based therapies. To demonstrate the utility of tumor-stroma proportion (TSP) as a predictive biomarker of chemoresistance of HGSC, progression-free survival (PFS), and overall survival (OS). This prognostic study leveraged tumors from patients with HGSC in The Cancer Genome Atlas (TCGA) cohort (1993-2013) and an independent cohort of resected clinical specimens from patients with HGSC (2004-2014) available in diagnostic and tissue microarray formats from the University of Tübingen in Germany. Data analysis was conducted from January 2021 to January 2024. Diagnosis of HGSC. Principal outcome measures were the ability of TSP to predict platinum chemoresistance, PFS, and OS. Using hematoxylin and eosin-stained slides from the Tübingen cohort (used for routine diagnostic assessment from surgical specimens) as well as tissue microarrays, representative sections of tumors for scoring of TSP were identified using previously evaluated cutoffs of 50% stroma or greater (high TSP) and less than 50% stroma (low TSP). Digitized slides from the TCGA Cohort were analyzed and scored in a similar fashion. Kaplan-Meier time-to-event functions were fit to estimate PFS and OS. The study included 103 patients (mean [SD] age, 61.6 [11.1] years) from the TCGA cohort and 192 patients (mean [SD] age at diagnosis, 63.7 [11.1] years) from the Tübingen cohort. In the TCGA cohort, there was no significant association of TSP levels with chemoresistance, PFS, or OS. However, in the Tübingen cohort, high TSP was associated with significantly shorter PFS (HR, 1.586; 95% CI, 1.093-2.302; P = .02) and OS (hazard ratio [HR], 1.867; 1.249-2.789; P = .002). Patients with chemoresistant tumors were twice as likely to have high TSP as compared to patients with chemosensitive tumors (HR, 2.861; 95% CI, 1.256-6.515; P = .01). In tissue microarrays from 185 patients from the Tübingen cohort, high TSP was again associated with significantly shorter PFS (HR, 1.675; 95% CI, 1.012-2.772 P = .04) and OS (HR, 2.491; 95% CI, 1.585-3.912; P < .001). In this prognostic study, TSP was a consistent and reproducible marker of clinical outcome measures of HGSC, including PFS, OS, and platinum chemoresistance. Accurate and cost-effective predictive biomarkers of platinum chemotherapy resistance are needed to identify patients most likely to benefit from standard treatments, and TSP can easily be implemented and integrated into prospective clinical trial design and adapted to identify patients who are least likely to benefit long-term from conventional platinum-based cytotoxic chemotherapy treatment at the time of initial diagnosis.

Analysis and Validation of Tyrosine Metabolism-related Prognostic Features for Liver Hepatocellular Carcinoma Therapy.

To explore tyrosine metabolism-related characteristics in liver hepatocellular carcinoma (LIHC) and to establish a risk signature for the prognostic prediction of LIHC. Novel prognostic signatures contribute to the mining of novel biomarkers, which are essential for the construction of a precision medicine system for LIHC and the improvement of survival. Tyrosine metabolism plays a critical role in the initiation and development of LIHC. Based on the tyrosine metabolism-related characteristics in LIHC, this study developed a risk signature to improve the prognostic prediction of patients with LIHC. To investigate the correlation between tyrosine metabolism and progression of LIHC and to develop a tyrosine metabolism-related prognostic model. Gene expression and clinicopathological information of LIHC were obtained from The Cancer Genome Atlas (TCGA) database. Distinct subtypes of LIHC were classified by performing consensus cluster analysis on the tyrosine metabolism-related genes. Univariate and Lasso Cox regression were used to develop a RiskScore prognosis model. Kaplan-Meier (KM) survival analysis with log-rank test and area under the curve (AUC) of receiver operating characteristic (ROC) were employed in the prognostic evaluation and prediction validation. Immune infiltration, tyrosine metabolism score, and pathway enrichment were evaluated using single-sample gene set enrichment analysis (ssGSEA). Finally, a nomogram model was developed with the RiskScore and other clinicopathological features. Based on the tyrosine metabolism genes in the TCGA cohort, we identified 3 tyrosine metabolism-related subtypes showing significant prognostic differences. Four candidate genes selected from the common differentially expressed genes (DEGs) between the 3 subtypes were used to develop a RiskScore model, which could effectively divide LIHC patients into high- and lowrisk groups. In both the training and validation sets, high-risk patients tended to have worse overall survival, less active immunotherapy response, higher immune infiltration and clinical grade, and higher oxidative, fatty, and xenobiotic metabolism pathways. Multivariate analysis confirmed that the RiskScore was an independent indicator for the prognosis of LIHC. The results from pan-- cancer analysis also supported that the RiskScore had a strong prognostic performance in other cancers. The nomogram demonstrated that the RiskScore contributed the most to the prediction of LIHC prognosis. Our study developed a tyrosine metabolism-related risk model that performed well in survival prediction, showing the potential to serve as an independent prognostic predictor for LIHC treatment.

Establishment of the molecular subtypes and a risk model for stomach adenocarcinoma based on genes related to reactive oxygen species

BackgroundOxidative stress promotes the development of stomach adenocarcinoma (STAD) and resistance of STAD patients to chemotherapy. This study developed a risk classification and prognostic model for STAD based on genes related to oxidative stress. MethodsUnivariate Cox regression and least absolute shrinkage and selection operator (Lasso) regression analysis were performed using transcriptome data of STAD from The Cancer Genome Atlas (TCGA) and reactive oxygen species (ROS)-related genes from Gene Set Enrichment Analysis (GSEA) website to develop a risk model. Genetic landscape, pathway characteristics and immune characteristics between the two risk groups were assessed to evaluate patients’ response to anti-tumor therapy. Further, a nomogram was created to evaluate the clinical outcomes of STAD patients. The mRNA levels of genes were detected by reverse transcription quantitative PCR (RT-qPCR). ResultsTwo ROS-related molecular subtypes (subtype C1 and C2) were classified, with subtype C2 having unfavorable prognosis, higher immune score, and greater infiltration of macrophages, myeloid-derived suppressor cells (MDSCs), mast cells, regulatory T cells, and C–C chemokine receptor (CCR). Five ROS-related genes (ASCL2, COMP, NOX1, PEG10, and VPREB3) were screened to develop a prognostic model, the robustness of which was validated in TCGA and external cohorts. RT-qPCR analysis showed that ASCL2, COMP, NOX1, and PEG10 were upregulated, while the mRNA level of VPREB3 was downregulated in gastric cancer cells. The risk score showed a negative relation to tumor mutation burden (TMB). Low-risk patients exhibited higher mutation frequencies of TTN, SYNE1, and ARID1A, higher response rate to immunotherapy and were more sensitive to 32 traditional chemotherapeutic drugs, while high-risk patients were sensitive to 13 drugs. Calibration curve and DCA confirmed the accuracy and reliability of the nomogram. ConclusionThese findings provided novel understanding on the mechanism of ROS in STAD. The current study developed a ROS-related signature to help predict the prognosis of patients suffering from STAD and to guide personalized treatment.