t-SNE Analysis Research Articles (Page 3)

Background: Drug-induced cardiotoxicity (DICT) is a severe adverse drug reaction that contributes to clinical trial failures and drug withdrawals. Forecasting DICT is challenging due to the property’s complex nature, and experimental assays are cumbersome and poorly correlate with human outcomes. Machine learning trained on clinical DICT data can quickly and accurately predict cardiotoxicity, pre-empting clinical failure and providing insight. Methods: We used our previously developed ADMET-AI machine learning platform to predict DICT and its sources. ADMET-AI is a freely available web tool (admet.ai.greenstonebio.com) that predicts 41 absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of a given molecule. We coupled ADMET-AI’s 41 ADMET property predictions with an Extreme Gradient Boosting (XGB) model to output a probability of cardiotoxicity. This XGB model was trained on 555 drugs from the published DICTrank dataset that have been labeled as having either no cardiotoxicity (262 drugs) or severe cardiotoxicity (293 drugs). Notably, over 67% of the DICTrank drugs are not in any of the 41 ADMET-AI training datasets. Results: The ADMET-AI-based cardiotoxicity XGB model, evaluated with 10-fold cross-validation, achieved an average area under the precision-recall curve (PR-AUC) of 0.75. Our model outperformed an XGB model trained on SwissADME predictions (PR-AUC=0.69). Shapley values identified CYP2D6 Substrate and Nrf2-Antioxidant Responsive Element as key predictors of severe cardiotoxicity, while CYP2D6 inhibition and Aromatase were linked to non-cardiotoxicity. t-SNE analysis of the DICTrank chemical space using the five most predictive ADMET-AI properties showed clear clustering of toxicity. Case studies of cisapride and lactulose illustrated the model’s ability to distinguish between cardiotoxic and non-cardiotoxic drugs based on predictions. Compared to publicly available models, ADMET-AI is currently the fastest and most accurate tool for ADMET and DICT prediction. Conclusions: ADMET-AI effectively predicts DICT with insight into potential causes, surpassing the current standard of public tools. The ADMET-AI-based cardiotoxicity prediction models are freely available to enable DICT prediction, with the potential to prevent late-phase clinical failures. While this study focuses on DICT, a similar workflow can adapt ADMET-AI to make predictions for other key properties in drug discovery.

Abstract INTRODUCTION Pilocytic astrocytoma (PA) is a circumscribed low-grade astrocytic glioma, generally considered to be associated with a good prognosis. However, some PA patients experience unfavorable outcomes; advanced age was known as an independent poor prognostic factor. This study retrospectively reviewed PA patients and conducted molecular analyses for elderly patients. METHODS We analyzed data from 29 histologically confirmed PA patients treated at a single center from 2002 to 2021 and conducted integrated molecular analyses including DNA methylation profiling for elderly PA patients. RESULTS The median age at diagnosis was 14 years (range 3-82 years), with 4 patients (14%) being elderly (patients≥60years old). Elderly patients had significantly worse progression-free survival and overall survival. DNA methylation analysis was performed on 2 of the 4 elderly patients. Despite histological diagnoses of PA, methylation profiling classified one case as high-grade astrocytoma with piloid features (methylation class scores below 0.3 in both v11b4 and v12.5) and the other as glioblastoma, IDH-wildtype (score over 0.5 in both v11b4 and v12.5), using classifiers from the German Cancer Research Center and t-SNE analysis. CONCLUSION Elderly patients with PA morphology showed unfavorable outcomes in this cohort. In those patients, DNA methylation profiling revealed the possibility of high-grade astrocytic tumors, including newly defined entities.

Abstract BACKGROUND Glioblastoma (GBM) is the most lethal primary malignant brain tumor with a median survival of 15–20 months. Concurrent temozolomide (TMZ) chemotherapy and radiation (XRT) remain the current standard of care (SOC) treatment for newly diagnosed GBM. The addition of tumor treating fields (TTFs) improves median survival for newly diagnosed GBM from 16 to 20.9 months. This modest improvement underscores the urgent need to identify a new treatment modality as a safe and effective therapy. PURPOSE The noninvasive Oncomagnetic device developed in our laboratory selectively kills glioblastoma (GBM) in vitro. We studied the cellular and molecular effects of Oncomagnetic monotherapy (OMT) on GBM cell lines and investigated immune response in a syngeneic mouse model. RESULTS OMT significantly reduces cell proliferation and cell survival in vitro. OMT induces persistent ROS primarily by increasing superoxide level in cancer cells and reduces mitochondrial-membrane potential in GBM cells. It also induces DNA damage and arrests cells in G1-phase of the cell cycle. Whole-body Oncomagnetic stimulation caused a substantial retardation of tumor growth and reduction of the contrast-enhanced tumor volume in 9.4T MRI scans. OMT showed significant increase in overall survival in treated mice. We also observed upregulated immune response indicated by RNA sequencing data obtained from OMT-treated GBM cells as well as in treated GBM mouse tumor sections by imaging mass cytometry analysis. The tSNE analysis of various immune clusters suggests significant increase in immune response by activating CD4+, CD8+, murine-macrophages and M1-macrophages located at tumor site in the treated group. CONCLUSION The obtained data indicate significant anti-tumor response by inhibiting cancer signaling pathways and an increased immune response. Our findings suggest unique mechanism of action for OMT stimulation and provide a strong rationale for standalone OMT for GBM.

Revolutionizing B-ALL Cell Line Development: Novel Generation to Uncover Therapeutic Vulnerabilities

To investigate the association between disulfidptosis-related genes (DFRGs) and patient prognosis, while concurrently identifying potential therapeutic targets in glioblastoma (GBM). We retrieved RNA sequencing data and clinical characteristics of GBM patients from the TCGA database. We found there was a total of 6 distinct clusters in GBM, which was identified by the t-SNE and UMAP dimension reduction analysis. Prognostically significant genes in GBM were identified using the limma package, coupled with univariate Cox regression analysis. Machine learning algorithms were then applied to identify central genes. The CIBERSORT algorithm was utilized to assess the immunological landscape across different GBM subtypes. In vitro and in vivo experiments were conducted to investigate the role of SPAG4 in regulating the proliferation, invasion of GBM, and its effects within the immune microenvironment. 23 genes, termed DFRGs, were successfully identified, demonstrating substantial potential for establishing a prognostic model for GBM. Single cell analysis revealed a significant correlation between DFRGs and the progression of GBM. Utilizing individual risk scores derived from this model enabled the stratification of patients into two distinct risk groups, revealing significant variations in immune infiltration patterns and responses to immunotherapy. Utilizing the random survival forests algorithm, SPAG4 was identified as the gene with the highest prognostic significance within our model. In vitro studies have elucidated SPAG4's significant role in GBM pathogenesis, potentially through the regulation of fatty acid metabolism pathways. Our in vivo investigations using a subcutaneous xenograft model have confirmed SPAG4's influence on tumor growth and its capacity to modulate the immune microenvironment. Advanced research hints that SPAG4 might achieve immune evasion by increasing CD47 expression, consequently reducing phagocytosis. These findings highlight SPAG4 as a potential GBM therapeutic target and emphasize the complexity of the immune microenvironment in GBM progression.

In response to the inadequacy of manual analysis in meeting the rising demand for retinal optical coherence tomography (OCT) images, a self-supervised learning-based clustering model was implemented. A public dataset was utilized, with 83,484 OCT images with categories of choroidal neovascularization (CNV), diabetic macular edema (DME), drusen, and normal fundus. This study employed the Semantic Pseudo Labeling for Image Clustering (SPICE) framework, a self-supervised learning-based method, to cluster unlabeled OCT images into binary and four categories, and the performances were compared with baseline models. We also analysed feature distribution using t-SNE, and explored the cluster centers, attention maps, and misclassified images. In addition, DME and CNV subsets were clustered binarily, and the results were interpreted by two retinal specialists. SPICE demonstrated superior performance in binary and four categories classification tasks, achieving the accuracy of 0.886 and 0.846, respectively. In t-SNE analysis, the four types exhibited significant clustering into distinct groups. The cluster centers corresponded to the human labels, and the heat map revealed that the model focused on important biomarkers. The misclassified images exposed similar features to the inaccurate classes. The model also grouped DME and CNV into two distinct categories respectively. Self-supervised clustering effectively distinguished disease variances and revealed common features, with a notable capability to detect disease heterogeneity through biomarkers.

Uterine mesenchymal tumours harboring KAT6B/A::KANSL1 gene fusions typically exhibit histological and immunophenotypic overlap with endometrial stromal and smooth muscle tumours. To date it remains uncertain whether such neoplasms should be regarded as variants of smooth muscle or endometrial stromal neoplasm, or if they constitute a distinct tumour type. In this study we investigated DNA methylation patterns and copy number variations (CNVs) in a series of uterine tumours harboring KAT6B/A::KANSL1 gene fusions in comparison to other mesenchymal neoplasms of the gynecological tract. Unsupervised hierarchical clustering and t-SNE analysis of DNA methylation data (Illumina EPIC array) identified a distinct cluster for 8/13 KAT6B/A::KANSL1 tumours (herein referred to as core cluster). The other 5 tumours (herein referred to as outliers) did not assign to the core cluster but clustered near various other tumour types. CNV analysis did not identify significant alterations in the core cluster. In contrast, various alterations, including deletions at the CDKN2A/B and NF1 loci were identified in the outlier group. Analysis of the DNA methylation clusters in relation to histological features revealed that while tumours in the core KAT6B/A::KANSL1 cluster were histologically bland, outlier tumours frequently exhibited “high-grade” histologic features in the form of significant nuclear atypia, increased mitotic activity and necrosis. Three of the five patients with outlier tumours died from their disease while clinical follow-up in the remaining two patients was limited (less than 12 months). In comparison, none of the 7 out of 8 patients with tumors in the core KAT6B/A::KANSL1 sarcoma cluster, where follow-up was available, died from disease. Furthermore, only 1 out of 7 patients recurred (mean follow-up of 30 months). In conclusion, KAT6B/A::KANSL1 uterine sarcoma is a molecularly unique type of uterine tumour that should be recognized as a distinct entity. These tumors typically exhibit low-grade histologic features but are occasionally morphologically high-grade; the latter have a DNA methylation profile different from the typical low-grade neoplasms and may be associated with aggressive behaviour.

iACP-DFSRA: Identification of Anticancer Peptides Based on a Dual-channel Fusion Strategy of ResCNN and Attention

Heavy metal contamination in food and water is a major public health concern because heavy metals are toxic in minute amounts. DNAzyme sensors are emerging as a promising tool for rapid onsite detection of heavy metals, which can aid in minimizing exposure. However, DNAzyme activity toward its target metal is not absolute and has cross-reactivity with similar metals, which is a major challenge in the wide-scale application of DNAzyme sensors for environmental monitoring. To address this, we constructed a four DNAzyme array (17E, GR-5, EtNA, and NaA43) and used a pattern-based readout to improve sensor accuracy. We measured cross-reactivity between three metal cofactors (Pb2+, Ca2+, and Na+) and common interferents (Mg2+, Zn2+, Mn2+, UO22+, Li+, K+, and Ag+) and then used t-SNE analysis to identify and quantify the metal ion. We further showed that this method can be used for distinguishing mixtures of metals and detecting Pb2+ in environmental soil samples at micromolar concentrations.

PURPOSEEpendymomas of the spinal cord are rare among children and adolescents, and the individual risk of disease progression is difficult to predict. Recent advances in subtyping the main molecular groups, myxopapillary ependymoma (MPE) and spinal ependymoma (SP-EPN), have successfully identified distinct risk groups among adult patients. This study aims at evaluating the prognostic impact of MPE subtypes A and B in a pediatric cohort. METHODSEighty-three patients with spinal ependymoma ≤22 years registered in the HIT-MED database between 1992 and 2022 were included. Forty-seven tumors were analyzed regarding global DNA methylation. In six cases, antibodies against HOXB13 and MYCN were used as surrogate markers for the methylation group. Twenty-eight MPE were further classified into subtypes by t-distributed stochastic neighbor embedding (t-SNE) analysis. RESULTSMPE (n=32, 63%) was the most common molecular type followed by SP-EPN (n=17, 33%) and MYCN-amplified ependymoma (n=2, 4%). One case remained molecularly unclear, and one was excluded due to reclassification as anaplastic pilocytic astrocytoma. MPE subtyping identified 18 MPE-A and 9 MPE-B (inconclusive: n=1). 5-year progression-free survival (5y-PFS) did not significantly differ between MPE and SP-EPN (65% vs. 78%, p=0.64). MYCN-amplification was associated with early relapses and with death in one patient. Patients with MPE-B may have a tendency towards superior 5y-PFS compared to MPE-A (86% vs. 56%, p=0.15), irrespective of the use of adjuvant treatment. MPE-A was more frequently disseminated at time of diagnosis (n=6/18) than MPE-B (n=1/8, not evaluable: n=1; p=0.28). CONCLUSIONMolecular subtyping of spinal ependymoma in children does not allow to separate patients into distinct risk groups, yet. However, the potentially favorable PFS among patients with MPE-B prompts the consideration of treatment de-intensification for this group. Larger cohorts and further insights into the molecular heterogeneity of these tumors are needed to complete the basis for future clinical decision-making.

Efficient multi-agent cooperation: Scalable reinforcement learning with heterogeneous graph networks and limited communication

Abstract Background and Aims Kidney transplantation (KTX) is the current treatment of choice in patients with kidney failure. Since immunosuppressive medication to prevent acute rejection is still needed in high doses with frequent adverse events, new strategies to achieve risk-appropriate personalization of immunosuppression are needed. Calcineurin inhibitors are the cornerstone of T cell suppression, currently available in different release formulations including once daily long-term release tacrolimus (LCPT) and regular twice daily immediate release tacrolimus (IRT). Little is known about the temporal evolution of tacrolimus metabolism potentially influencing T cell composition and clinical endpoints between different tacrolimus formulations after KTX. Method To evaluate the influence of de novo LCPT on clinical endpoints, metabolizer phenotype and immune cell composition after KTX, subgroup analysis of a single-center, prospective, observational cohort study including 32 patients started on LCPT (0.1 mg/kg) and 55 patients started on IRT (0.1 mg/kg) was conducted. Peripheral blood mononuclear cells were isolated from whole blood samples drawn before KTX, 2 months and 12 months after KTX for flow cytometry analysis. Study visits were simultaneously conducted to obtain tacrolimus dosing, plasma levels, acute rejection and infection incidence. Statistical analysis was done using GraphPad Prism (v10), accepting significance for p < 0.05, and flow cytometry analysis was done using FlowJo (v10) software. Results The cohort represented KTX recipients with a median age of 60 years and similar distributions of gender, HLA-mismatches and induction therapy. Clinical outcome data showed a trend towards lower acute rejection rate, lower serum-creatinine values and increased incidence of BK-viremia in LCPT treated patients. Rejection rate was associated with a fast-metabolizer phenotype (HR 3.9; 1.1-6.8, p < 0.01) and HLA mismatch > 3 (HR 4.0; 1.2-7.3, p < 0.01) while this effect was blunted in the LCPT group (HR 1.9; 0.7-3.2, p = 0.67 and HR 0.9; 0.3-1.4, p = 0.38 respectively) in a multivariate Cox-regression model. Cluster exploration and tSNE analysis of flow cytometry data revealed diminution of activated-proliferative effector and regulatory T cell trajectories, with significant reduction in cell counts of CD4+CD15s+/CD45RA-Ki67+Foxp3+ Treg and in CD4+CD147high/CD45RA-LAP+HLADR+FcRL3+ effector T cells in LCPT treated patients after KTX. Conclusion Induction with de novo LCPT was associated with a blunted influence of metabolizer phenotype and HLA-mismatch on acute rejection in our cohort. Major changes were detected in peripheral T cell composition after transplantation, with signs of a stronger immunosuppressive effect of LCPT on circulating activated T cells. Therefore, LCPT may be considered for KTX recipients with higher HLA-mismatch (i.e. “old-for-old”) or fast-metabolizer phenotype. Inclusion of changes in T cell composition as biologic effect measures may be worthwhile in future studies evaluating biomarkers to guide immunosuppression and predict complications after KTX.

Generative AI (GenAI) and LLMs have started to influence how teachers teach and studentslearn, including the ones in programming languages and tutoring. However, there have been debates onwhether AI could be beneficial to students’ learning or not. One way to see this issue is from the perspectives of thestudents. Therefore, this study explored how students perceive the use of AI in their education. The data wasgathered through interviews with 62 students and other stakeholders, such as instructors and IT specialists. The results showed that the students positively perceived using AI as a tutor. Moreover, this study suggests several things to consider when integrating AI tutors for programming. The findings reveal positive student perceptions regarding AI's potential within the teaching-learning process. Students envision AI tutors offering personalized assistance, adapting to individual learning styles, and providing immediate feedback, potentially augmenting their grasp of programming concepts. We applied Statistical analysis, machine learning, and natural language processing techniques such as PCA, t-SNE, LDA, and sentiment analysis.

Cellular myxoma is a benign soft tissue tumor frequently associated with GNAS mutation that may morphologically resemble low-grade myxofibrosarcoma. This study aimed to identify the undescribed methylation profile of cellular myxoma and compare it to myxofibrosarcoma. We performed molecular analysis on twenty cellular myxomas and nine myxofibrosarcomas and analyzed the results using the methylation-based DKFZ sarcoma classifier. A total of 90% of the cellular myxomas had GNAS mutations (four loci had not been previously described). Copy number variations were found in all myxofibrosarcomas but in none of the cellular myxomas. In the classifier, none of the cellular myxomas reached the 0.9 threshold. Unsupervised t-SNE analysis demonstrated that cellular myxomas form their own clusters, distinct from myxofibrosarcomas. Our study shows the diagnostic potential and the limitations of molecular analysis in cases where morphology and immunohistochemistry are not sufficient to distinguish cellular myxoma from myxofibrosarcoma, particularly regarding GNAS wild-type tumors. The DKFZ sarcoma classifier only provided a valid prediction for one myxofibrosarcoma case; this limitation could be improved by training the tool with a more considerable number of cases. Additionally, the classifier should be introduced to a broader spectrum of mesenchymal neoplasms, including benign tumors like cellular myxoma, whose distinct methylation pattern we demonstrated.

This study aimed to automatically detect epiretinal membranes (ERM) in various OCT-scans of the central and paracentral macula region and classify them by size using deep-neural-networks (DNNs). To this end, 11,061 OCT-images were included and graded according to the presence of an ERM and its size (small 100–1000 µm, large > 1000 µm). The data set was divided into training, validation and test sets (75%, 10%, 15% of the data, respectively). An ensemble of DNNs was trained and saliency maps were generated using Guided-Backprob. OCT-scans were also transformed into a one-dimensional-value using t-SNE analysis. The DNNs’ receiver-operating-characteristics on the test set showed a high performance for no-ERM, small-ERM and large-ERM cases (AUC: 0.99, 0.92, 0.99, respectively; 3-way accuracy: 89%), with small-ERMs being the most difficult ones to detect. t-SNE analysis sorted cases by size and, in particular, revealed increased classification uncertainty at the transitions between groups. Saliency maps reliably highlighted ERM, regardless of the presence of other OCT features (i.e. retinal-thickening, intraretinal pseudo-cysts, epiretinal-proliferation) and entities such as ERM-retinoschisis, macular-pseudohole and lamellar-macular-hole. This study showed therefore that DNNs can reliably detect and grade ERMs according to their size not only in the fovea but also in the paracentral region. This is also achieved in cases of hard-to-detect, small-ERMs. In addition, the generated saliency maps can be used to highlight small-ERMs that might otherwise be missed. The proposed model could be used for screening-programs or decision-support-systems in the future.

Colorectal cancer (CRC) is characterized by a high metastasis rate, leading to poor prognosis and increased mortality. Anoikis, a physiological process, serves as a crucial barrier against metastasis. The objective of this research is to construct a prognostic model for CRC based on genes associated with anoikis. The study involved differential analysis and univariate Cox analysis of anoikis-related genes (ARGs), resulting in the selection of 47 genes closely associated with prognosis. Subsequently, unsupervised k-means clustering analysis was conducted on all patients to identify distinct clusters. Survival analysis, principal component analysis (PCA), and t-distributed stochastic neighbor embedding (t-SNE) analysis were performed on the different clusters to investigate associations within the clusters. Gene set variation analysis (GSVA) and gene set enrichment analysis (GSEA) were utilized to assess metabolic pathway enrichment between the identified clusters. Furthermore, single-sample GSEA (ssGSEA) was applied to explore variations in immune infiltration. Multivariable Cox regression and least absolute shrinkage and selection operator (LASSO) analyses were conducted to construct a risk model based on ten signatures, which enabled the grouping of all samples according to their risk scores. The prognostic value of the model was validated using receiver operating characteristic (ROC) curves, area under the curve (AUC) calculations, and survival curves. Additionally, the expression of candidate genes was validated using quantitative real-time polymerase chain reaction (qRT-PCR). Forty-seven survival-related ARGs were screened out. Somatic mutation analysis showed that these genes revealed a high mutation rate. Based on their expression, two clusters were identified. Cluster B patients exhibited a shortened overall survival and higher immune infiltration. A risk scoring model including ten genes was subsequently developed, which exhibited excellent prognostic predictive ability for CRC, as evidenced by the survival curve, ROC curve, and AUC curve. In addition, a nomogram was developed for predicting 3- and 5-year survival probabilities. The qRT-PCR results indicated the dissimilarities among the ten signatures in the tumor tissues and adjacent tissues of patients with CRC were fundamentally consistent with the analytical findings. This study comprehensively evaluated the prognostic significance of ARGs in CRC. It identified two distinct anoikis-related clusters and examined their respective immune microenvironments. Furthermore, an ARGs signature was developed to effectively predict the prognosis of CRC, thereby establishing a solid foundation for investigating the clinical prognostic role of anoikis in CRC.

Abstract Background: Cell-free DNA (cfDNA) detected in proximal body fluids has demonstrated potential for cancer detection using minimally invasive methodology. Our past work showed that tumor cfDNA is present in the cerebrospinal fluid (CSF) and other body fluids of patients with inconclusive standard of care testing. However, past work measuring copy number aberrations or somatic mutations was limited in cancer classification. To facilitate reliable classification, even at low tumor fractions and with fragmented DNA, we developed XR-methylSeq, a methylation sequencing platform to enrich for cell type-specific markers. Methods: We benchmarked XR-methylSeq with the K562 cell line and correlated the methylation values with gold standard measurement − whole genome bisulfite sequencing (WGBS). Methylation classifiers were applied for at least 22 cytology-positive body fluids, incorporating methylation array data from public references. T-distributed stochastic neighbor embedding (t-SNE) analysis was used for visualization in R. Deconvolution of cell type fractions for at least 29 (seven cytology-negative) body fluids and plasma samples was conducted using wgbstools. Cell type-specific markers were identified from a human DNA methylation atlas. Results: Benchmarks: XR-methylSeq has a 5-fold enrichment of the cell type-specific markers compared with WGBS. XR-methylSeq at 20 ng input highly correlates with WGBS at 2 μg (Pearson’s r = 0.97). Body Fluids: Thirteen cytology-positive CSF samples had copy number aberrations, 77% of them had concordant tumor classification, while the remaining 23% clustered with low tumor fraction samples. All nine lung primaries, including a low tumor fraction case that did not originally classify, showed a consistent cell-of-origin through deconvolution, as indicated by increased contributions from lung alveolar epithelial cells. Among six other body fluids, three exhibited the highest fractions aligning with the clinically identified cancer cell-of-origin. Additionally, the plasma cfDNA of a patient with acute liver injury had a higher fraction of hepatocyte signatures (22%) than the healthy control (8%). Conclusions: This research highlights the potential of XR-methylSeq as an enriched methylation profiling method useful for liquid biopsy applications. Citation Format: Jingru Yu, Lauren S. Ahmann, Yvette Y. Yao, Angus Toland, Alicia Snowden, Chandler Ho, Benjamin Pinsky, Hannes Vogel, Ruben Y. Luo, Linlin Wang, Brooke Howitt, Brittany Holmes, Alarice C. Lowe, Wei Gu. Tumor classification and deconvolution in liquid biopsy using enriched methylation sequencing [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 7565.

Abstract Diffuse midline gliomas (DMG) H3 K27-altered are incurable grade 4 gliomas and represent a major challenge in neuro-oncology. This tumor type is now classified in four subtypes by the 2021 edition of the WHO Classification of the Central Nervous System (CNS) tumors. However, the H3.3-K27M subgroup still appears clinically and molecularly heterogenous. Recent publications reported that rare patients presenting a co-occurrence of H3.3K27M with BRAF or FGFR1 alterations tended to have a better prognosis. To better study the role of these co-driver alterations, we assembled a large pediatric and adult cohort of 29 tumors H3K27-altered with co-occurring activating mutation in BRAF or FGFR1 as well as 31 previous cases from the literature. We performed a comprehensive histological, radiological, genomic, transcriptomic and DNA methylation analysis. Interestingly, unsupervised t-distributed Stochastic Neighbor Embedding (tSNE) analysis of DNA methylation profiles regrouped BRAFV600E and all but one FGFR1MUT DMG in a unique methylation cluster, distinct from the other DMG subgroups and also from ganglioglioma (GG) or high-grade astrocytoma with piloid features (HGAP). This new DMG subtype harbors atypical radiological and histopathological profiles with calcification and/or a solid tumor component both for BRAFV600E and FGFR1MUT cases. Contrary to other DMG, these tumors occur more frequently in the thalamus (70% for BRAFV600E and 58% for FGFR1MUT) and patients have a longer overall survival with a median above three years. The analyses of a H3.3-K27M BRAFV600E tumor at diagnosis and corresponding in vitro cellular model showed that mutation in H3-3A was the first event in the oncogenesis, indicating a distinct oncogenesis from classical DMG despite an identical founder H3K27 alteration, and possible subclonal evolution.In conclusion, DMG, H3 K27 and BRAF/FGFR1 co-altered represent a new subtype of DMG with distinct genotype-phenotype characteristics, which deserve further attention with respect to trial interpretation and patient management. Our work also lays the foundations for therapeutic development highly-needed against this deadly disease. Citation Format: Lucie Auffret, Yassine Ajlil, Arnault Tauziede-Espariat, Thomas Kergrohen, Chloé Puiseux, Laurent Riffaud, Pascale Blouin, Anne-Isabelle Bertozzi, Pierre Leblond, Klas Blomgren, Sebastien Froelich, Alberto Picca, Mehdi Touat, Marc Sanson, Kevin Beccaria, Thomas Blauwblomme, Volodia Dangouloff-Ros, Nathalie Boddaert, Pascale Varlet, Marie-Anne Debily, Jacques Grill, David Castel. Clinico-radiological and histomolecular analyses identify of a new subtype of diffuse midline glioma, H3 K27 and BRAF/FGFR1 co-altered [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 155.

Abstract Ultraconserved regions (UCRs), encompassing 481 genomic segments identical across humans, mice, and rats, and conserved in other species, are distributed across all human chromosomes except 21 and Y. Over 90% of these regions, termed transcribed ultraconserved regions (T-UCRs) are expressed in at least one human tissue. Although extensively studied in the context of cancer since their discovery, the prognostic significance of T-UCRs in a pan-cancer context has not been explored. For this, TCGA data extracted from the TANRIC platform was utilized, differential expression and survival analysis was performed on different tumor types. Our analysis revealed that approximately 300 UCRs were expressed in the majority of the 20 tumor tissues. Additionally, it was noted that all examined tumor tissues expressed a higher number of UCRs compared to their non-tumor counterparts. Through t-distributed stochastic neighbor embedding (t-SNE) analysis, UCR expression patterns could distinguish primary tumor origins and types. Regarding the prognostic value, we identified 100 (20,8%) T-UCRs associated with disease-specific survival (DSS) and 102 (21.2%) T-UCRs associated with progression-free interval (PFI). Those T-UCRs are particularly important in the survival outcomes of kidney renal clear cell carcinoma (KIRC) and low-grade gliomas (LGC). In addition, we verified that uc.44, uc.48, uc.135, uc.144, uc.153, uc.217, uc.255, uc.256, uc.344, uc.357, uc.390, uc.427 and uc.436 were associated with survival of more than one tumor. Particularly, uc.135 high expression was associated with poor prognosis in Kidney Renal Papillary Cell Carcinoma (KIRP) and LGC, and good prognosis in KIRC. Uc.135 are mapped on MECOM gene, coding an oncoprotein acting as transcriptional regulator and involved in apoptosis, proliferation, and cell differentiation. Our findings suggest that analysis of T-UCRs expression have the potential to be used as predictive biomarkers and can help in studying the mechanisms and specific roles of the highlighted T-UCRs in cancer. Citation Format: Ana Carolina Rodrigues, Douglas Adamoski, Daniela F. Gradia, Jaqueline Carvalho de Oliveira. Transcribed ultraconserved regions are associated with survival in multiple tumor types [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 3681.

Abstract DNA methylation has been extensively adopted for the classification of brain tumors and is endorsed by current neuropathology guidelines. Recently, oligosarcoma, IDH-mutant has been characterized as a distinct group of IDH mutant gliomas by DNA methylation classification. However, emerging studies revealed the variety in this group. Here, we employed DNA methylation-based t-SNE analysis to characterize oligosarcomas identified by the dkfz brain tumor classifier (v12.5). Thereby, we uncovered that oligosarcomas were segregated into two distinct groups, namely, subgroup A and subgroup B, which were further validated by genetic alterations and clinical outcomes. Next-generation Sequencing revealed subtype A tumors enriched with chr 1p/19q co-deletion, chr 6q loss, CDKN2A/B homogeneous deletion, TERT promoter mutation, and NF1 mutation, while subtype B tumors showed chr7 gaining and TP53 mutation. RNA sequencing and gene set enrichment analysis (GSEA) revealed that neuronal-tumor interaction and muscular progression pathways were more enriched in subtype A tumors. In contrast, subtype B tumors showed enrichment for angiogenesis and mTOR pathways. Additionally, inter-chromosomal fusion genes were more frequently found in subtype A tumors. Clinically, patients with subgroup A oligosarcomas manifested poorer survival and distinct imaging features in magnetic resonance imaging (MRI) compared to those with subgroup B. Collectively, we provided more cases of oligosarcoma, IDH-mutant, confirming the existence of this newly identified group, and revealed the subgroups harboring different genetic alterations and suffering different prognoses, providing new evidence for the molecular pathology in diffuse glioma. Citation Format: Kenan Zhang, Lingyu Liu, Xing Liu, Tao Jiang. DNA methylation profiling identifies two distinct groups of oligosarcoma, IDH-mutant with distinct genetic alterations and clinical outcomes [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 7564.