High Intratumoral Heterogeneity Research Articles

Mouse models of pediatric high-grade gliomas with MYCN amplification reveal intratumoral heterogeneity and lineage signatures

Pediatric high-grade gliomas of the subclass MYCN (HGG-MYCN) are highly aggressive tumors frequently carrying MYCN amplifications, TP53 mutations, or both alterations. Due to their rarity, such tumors have only recently been identified as a distinct entity, and biological as well as clinical characteristics have not been addressed specifically. To gain insights into tumorigenesis and molecular profiles of these tumors, and to ultimately suggest alternative treatment options, we generated a genetically engineered mouse model by breeding hGFAP-cre::Trp53Fl/Fl::lsl-MYCN mice. All mice developed aggressive forebrain tumors early in their lifetime that mimic human HGG-MYCN regarding histology, DNA methylation, and gene expression. Single-cell RNA sequencing revealed a high intratumoral heterogeneity with neuronal and oligodendroglial lineage signatures. High-throughput drug screening using both mouse and human tumor cells finally indicated high efficacy of Doxorubicin, Irinotecan, and Etoposide as possible therapy options that children with HGG-MYCN might benefit from.

The loss of neoantigens is an important reason for immune escape in multiple myeloma patients with high intratumor heterogeneity.

Intratumor heterogeneity (ITH) is an important factor for clinical outcomes in patients with multiple myeloma (MM). High ITH has been proven to be a key reason for tumor immune escape and treatment resistance. Neoantigens are thought to be associated with ITH, but the specific correlation and functional basis for this remains unclear. We study this question through the whole-exome sequencing (WES) data from 43 high ITH newly diagnosed MM patients in our center. Mutant allele tumor heterogeneity (MATH) was conducted to quantify ITH. The cutoff value for high intratumor heterogeneity was determined by comparing MATH of different kinds of tumors. NeoPredPipe was performed to predict neoantigens and binding affinity. Compared to other tumors, MM has a relatively low tumor mutation burden but a high ITH. Patients with high MATH had significantly shorter progression-free survival times than those with low MATH (p = 0.001). In high ITH samples, there is a decrease in strong-binding neoantigens (p = 0.019). The loss of strong-binding neoantigens is a key factor for insensitivity to therapy (p = 0.015). Loss of heterozygosity in HLA was not observed. In addition, patients with fewer neoantigens loss had higher rates of disease remission (p = 0.047). CD8 + T cells (p = 0.012) and NK cells (p = 0.011) decreased significantly in patients with high neoantigens loss rate. A prediction model based on neoantigens was built to evaluate the strength of immune escape. The loss of strong-binding neoantigens explains why tumors with high ITH have a higher degree of immune escape and may be feasible for deciding the clinical treatment of MM.

CSIG-15. PTN-PTPRZ1 SIGNALING MEDIATES TUMOR-NORMAL CROSSTALK IN GLIOBLASTOMA

Abstract Glioblastoma (GBM) is the most devastating form of brain cancer with poor patient prognosis and high recurrence. An increasing body of literature suggests crosstalk between tumor and its surroundings, both molding the immune microenvironment and forming functional synapses with neighboring normal cells. Despite the high intra-tumoral and inter-patient heterogeneity, we have discovered PTN-PTPRZ1 signaling as the most significant and preserved communication pathway between GBM cells and their immediate neighboring cells. Through a novel tumor transplantation protocol onto cortical organoids, which are human embryonic stem cell-derived aggregates that faithfully mimic the human brain, we have seen evidence of tumor-normal crosstalk where PTN-PTPRZ1 is a key signaling axis. We hypothesize that PTN-PTPRZ1 signaling is pivotal to tumor growth and invasion, which ultimately leads to recurrence. Modulating PTN and/or PTPRZ1 levels on the cortical organoids via shRNA affects PTN and/or PTPRZ1 expression on the tumors, suggesting dynamic communication between tumor and normal cells via PTN-PTPRZ1 signaling. During the process of generating PTN knockdown cortical organoids, we have additionally discovered a pivotal role for PTN in the very early stages of cortical development. PTN and PTPRZ1 are signature markers of the outer radial glia in the developing human brain, which are absent in mouse models where the majority of PTN biology has been studied. By supplementing exogenous PTN to PTN knockdown organoids and withdrawing it at various timepoints, we have identified drastic phenotypes of neuron and radial glia survival and cell fate specification. With cortical organoids, we can interrogate the role of PTN in the early stages of neurodevelopment via a human-derived system.

Successes and challenges in modeling heterogeneous BRAFV600E mutated central nervous system neoplasms.

Central nervous system (CNS) neoplasms are difficult to treat due to their sensitive location. Over the past two decades, the availability of patient tumor materials facilitated large scale genomic and epigenomic profiling studies, which have resulted in detailed insights into the molecular underpinnings of CNS tumorigenesis. Based on results from these studies, CNS tumors have high molecular and cellular intra-tumoral and inter-tumoral heterogeneity. CNS cancer models have yet to reflect the broad diversity of CNS tumors and patients and the lack of such faithful cancer models represents a major bottleneck to urgently needed innovations in CNS cancer treatment. Pediatric cancer model development is lagging behind adult tumor model development, which is why we focus this review on CNS tumors mutated for BRAFV600E which are more prevalent in the pediatric patient population. BRAFV600E-mutated CNS tumors exhibit high inter-tumoral heterogeneity, encompassing clinically and histopathological diverse tumor types. Moreover, BRAFV600E is the second most common alteration in pediatric low-grade CNS tumors, and low-grade tumors are notoriously difficult to recapitulate in vitro and in vivo. Although the mutation predominates in low-grade CNS tumors, when combined with other mutations, most commonly CDKN2A deletion, BRAFV600E-mutated CNS tumors are prone to develop high-grade features, and therefore BRAFV600E-mutated CNS are a paradigm for tumor progression. Here, we describe existing in vitro and in vivo models of BRAFV600E-mutated CNS tumors, including patient-derived cell lines, patient-derived xenografts, syngeneic models, and genetically engineered mouse models, along with their advantages and shortcomings. We discuss which research gaps each model might be best suited to answer, and identify those areas in model development that need to be strengthened further. We highlight areas of potential research focus that will lead to the heightened predictive capacity of preclinical studies, allow for appropriate validation, and ultimately improve the success of "bench to bedside" translational research.

Overview of Tumor Heterogeneity in High-Grade Serous Ovarian Cancers.

Ovarian cancers encompass a group of neoplasms originating from germinal tissues and exhibiting distinct clinical, pathological, and molecular features. Among these, epithelial ovarian cancers (EOCs) are the most prevalent, comprising five distinct tumor histotypes. Notably, high-grade serous ovarian cancers (HGSOCs) represent the majority, accounting for over 70% of EOC cases. Due to their silent and asymptomatic behavior, HGSOCs are generally diagnosed in advanced stages with an evolved and complex genomic state, characterized by high intratumor heterogeneity (ITH) due to chromosomal instability that distinguishes HGSOCs. Histologically, these cancers exhibit significant morphological diversity both within and between tumors. The histologic patterns associated with solid, endometrioid, and transitional (SET) and classic subtypes of HGSOCs offer prognostic insights and may indicate specific molecular profiles. The evolution of HGSOC from primary to metastasis is typically characterized by clonal ITH, involving shared or divergent mutations in neoplastic sub-clones within primary and metastatic sites. Disease progression and therapy resistance are also influenced by non-clonal ITH, related to interactions with the tumor microenvironment and further genomic changes. Notably, significant alterations occur in nonmalignant cells, including cancer-associated fibroblast and immune cells, during tumor progression. This review provides an overview of the complex nature of HGSOC, encompassing its various aspects of intratumor heterogeneity, histological patterns, and its dynamic evolution during progression and therapy resistance.

Comprehensive genomic characterization of sporadic synchronous colorectal cancer: Implications for treatment optimization and clinical outcome

Sporadic synchronous colorectal cancer (SCRC) refers to multiple primary CRC tumors detected simultaneously in an individual without predisposing hereditary conditions, which accounts for the majority of multiple CRCs while lacking a profound understanding of the genomic landscape and evolutionary dynamics to optimize its treatment. In this study, 103 primary tumor samples from 51 patients with SCRC undergo whole-exome sequencing. The germline and somatic mutations and evolutionary and clinical features are comprehensively investigated. Somatic genetic events are largely inconsistent between paired tumors. Compared with solitary CRC, SCRCs have higher prevalence of tumor mutation burden high (TMB-H; 33.3%) and microsatellite-instability high (MSI-H; 29.4%) and different mutation frequencies in oncogenic signaling pathways. Moreover, neutrally evolving SCRC tumors are associated with higher intratumoral heterogeneity and better prognosis. These findings unveil special molecular features, carcinogenesis, and prognosis of sporadic SCRC. Strategies for targeted therapy and immunotherapy should be optimized accordingly.

JS06.4.A MULTI-OMIC CHARACTERIZATION OF LUNG CANCER BRAIN METASTASIS REVEALS THE METABOLIC VULNERABILITY AS A NOVEL THERAPEUTIC TARGET

Abstract Lung cancer brain metastases (LC-BrMs) are frequently associated with dismal mortality rates in patients with lung cancer; however, standard of care therapies for LC-BrMs are still limited in their efficacy. A better understanding of the molecular mechanisms and tumor microenvironment of LC-BrMs will provide us with new insights into developing novel therapeutics for treating patients with LC-BrMs. Here, we performed integrated analyses of both bulk and single cell genomic transcriptomic and proteomic data derived from five published and two newly generated patient cohorts, totaling 141 patients with paired and unpaired primary lung cancer and LC-BrM. This was subsequently validated by immunohistochemistry (IHC) and multiplexed immunofluorescence (mIF) staining. LC-BrMs exhibited significantly lower tumor mutational burden but higher intra-tumor heterogeneity. We observed that mutations in a subset of genes were almost always shared by both primary lung cancers and LC-BrM lesions, including TP53, MUC16, LRP1B, KRAS, FAT3, and NF1. In addition, the genome-wide landscape of somatic copy number alterations (SCNAs) was found to be similar between primary lung cancers and LC-BrM lesions. Nevertheless, three regions of focal amplification were significantly enriched in LC-BrMs, including 11q13.1, 14q11.2, and 12p13.31. Integrated analyses of transcriptional and proteomic data revealed mitochondrial-specific metabolism was activated in LC-BrMs and accompanied by a suppressed tumor immune microenvironment, which was further corroborated by IHC and mIF data. Importantly, targeting oxidative phosphorylation with gamitrinib in combination with anti-PD1 immune checkpoint blockade (ICB) therapy significantly improved the survival of mice with LC-BrMs. In conclusion, our findings not only provide comprehensive and multi-omic perspectives of the molecular mechanisms of LC-BrMs but also contribute to the development of potential therapeutic strategies with the goal of translating it into clinical trials for patients with LC-BrMs.

P02.05.A DISSECTING THE DIFFUSE MIDLINE GLIOMA TUMOUR MICROENVIRONMENT COMMUNICATIONS USING MULTI-OMICS APPROACHES

Abstract BACKGROUND Diffuse midline glioma (DMG) represents a highly aggressive pediatric brain tumour with no chance of survival. Treatment options are urgently needed, and efforts have been put towards the development of T cell immunotherapy. However, this should consider the unique tumour microenvironment (TME) of DMG, both in terms of anatomical location and developmental stage. MATERIAL AND METHODS To gain knowledge on this unique tumour niche, we implemented an immuno-competent, syngeneic mouse model based on in-utero electroporation (IUE) of the key driver-mutations to induce tumour formation at embryonic stage. By combining snRNAseq, Spatial Transcriptomics and state-of-the-art imaging, we characterized the spatial and molecular landscape of DMG. RESULTS AND CONCLUSION We observed a high intratumoral heterogeneity with spatially restricted cell populations. Importantly, their molecular profiles correlate with patient data, illustrating the clinical relevance of the IUE model. Analysis of cell-cell communications in the TME revealed a novel signalling between DMG and immunosuppressive myeloid cells, that could impact T-cell response. Using an advanced patient-derived DMG organoid co-culture model with macrophages and engineered T-cells, we are currently investigating the role of this pathway on T-cells mediated killing of DMG organoids. Thus, this project fills a critical need to delineate and overcome the immune-suppressive environment in DMG and potentiate T cell targeting.

Convergent insights into intratumor heterogeneity

Mathematics, conventional histology, and genomics converge to confirm that highly aggressive clear cell renal cell carcinomas (CCRCCs) display low levels of intratumor heterogeneity (ITH). We hypothesize that therapeutic strategies aimed at maintaining high ITH levels would be advisable to slow down cancer evolution and to improve survival.

Cell-in-Cell–Mediated Entosis Reveals a Progressive Mechanism in Pancreatic Cancer

Pancreatic ductal adenocarcinoma (PDAC) is a deadly malignancy with high intratumoral heterogeneity. There is a lack of effective therapeutics for PDAC. Entosis, a form of nonapoptotic regulated cell death mediated by cell-in-cell structures (CICs), has been reported in multiple cancers. However, the role of entosis in PDAC progression remains unclear. CICs were evaluated using immunohistochemistry and immunofluorescence staining. The formation of CICs was induced by suspension culture. Through fluorescence-activated cell sorting and single-cell RNA sequencing, entosis-forming cells were collected and their differential gene expression was analyzed. Cell functional assays and mouse models were used to investigate malignant phenotypes. Clinical correlations between entosis and PDAC were established by retrospective analysis. Entosis was associated with an unfavorable prognosis for patients with PDAC and was more prevalent in liver metastases than in primary tumors. The single-cell RNA sequencing results revealed that several oncogenes were up-regulated in entosis-forming cells compared with parental cells. These highly entotic cells demonstrated higher oncogenic characteristics invitro and invivo. NET1, neuroepithelial cell transforming gene 1, is an entosis-related gene that plays a pivotal role in PDAC progression and is correlated with poor outcomes. Entosis is correlated with PDAC progression, especially in liver metastasis. NET1 is a newly validated entosis-related gene and a molecular marker of poor outcomes. PDAC cells generate a highly aggressive subpopulation marked by up-regulated NET1 via entosis, which may drive PDAC progression.

Decoupled neoantigen cross-presentation by dendritic cells limits anti-tumor immunity against tumors with heterogeneous neoantigen expression.

Cancer immunotherapies, in particular checkpoint blockade immunotherapy (CBT), can induce control of cancer growth, with a fraction of patients experiencing durable responses. However, the majority of patients currently do not respond to CBT and the molecular determinants of resistance have not been fully elucidated. Mounting clinical evidence suggests that the clonal status of neoantigens (NeoAg) impacts the anti-tumor T cell response. High intratumor heterogeneity (ITH), where the majority of NeoAgs are expressed subclonally, is correlated with poor clinical response to CBT and poor infiltration with tumor-reactive T cells. However, the mechanism by which ITH blunts tumor-reactive T cells is unclear. We developed a transplantable murine lung cancer model to characterize the immune response against a defined set of NeoAgs expressed either clonally or subclonally to model low or high ITH, respectively. Here we show that clonal expression of a weakly immunogenic NeoAg with a relatively strong NeoAg increased the immunogenicity of tumors with low but not high ITH. Mechanistically we determined that clonal NeoAg expression allowed cross-presenting dendritic cells to acquire and present both NeoAgs. Dual NeoAg presentation by dendritic cells was associated with a more mature DC phenotype and a higher stimulatory capacity. These data suggest that clonal NeoAg expression can induce more potent anti-tumor responses due to more stimulatory dendritic cell:T cell interactions. Therapeutic vaccination targeting subclonally expressed NeoAgs could be used to boost anti-tumor T cell responses.

Deep learning radiopathomics based on preoperative US images and biopsy whole slide images can distinguish between luminal and non-luminal tumors in early-stage breast cancers.

For patients with early-stage breast cancers, neoadjuvant treatment is recommended for non-luminal tumors instead of luminal tumors. Preoperative distinguish between luminal and non-luminal cancers at early stages will facilitate treatment decisions making. However, the molecular immunohistochemical subtypes based on biopsy specimens are not always consistent with final results based on surgical specimens due to the high intra-tumoral heterogeneity. Given that, we aimed to develop and validate a deep learning radiopathomics (DLRP) model to preoperatively distinguish between luminal and non-luminal breast cancers at early stages based on preoperative ultrasound (US) images, and hematoxylin and eosin (H&E)-stained biopsy slides. This multicentre study included three cohorts from a prospective study conducted by our team and registered on the Chinese Clinical Trial Registry (ChiCTR1900027497). Between January 2019 and August 2021, 1809 US images and 603H&E-stained whole slide images (WSIs) from 603 patients with early-stage breast cancers were obtained. A Resnet18 model pre-trained on ImageNet and a multi-instance learning based attention model were used to extract the features of US and WSIs, respectively. An US-guided Co-Attention module (UCA) was designed for feature fusion of US and WSIs. The DLRP model was constructed based on these three feature sets including deep learning US feature, deep learning WSIs feature and UCA-fused feature from a training cohort (1467 US images and 489 WSIs from 489 patients). The DLRP model's diagnostic performance was validated in an internal validation cohort (342 US images and 114 WSIs from 114 patients) and an external test cohort (279 US images and 90 WSIs from 90 patients). We also compared diagnostic efficacy of the DLRP model with that of deep learning radiomics model and deep learning pathomics model in the external test cohort. The DLRP yielded high performance with area under the curve (AUC) values of 0.929 (95% CI 0.865-0.968) in the internal validation cohort, and 0.900 (95% CI 0.816-0.953) in the external test cohort. The DLRP also outperformed deep learning radiomics model based on US images only (AUC 0.815 [0.719-0.889], p<0.027) and deep learning pathomics model based on WSIs only (AUC 0.802 [0.704-0.878], p<0.013) in the external test cohort. The DLRP can effectively distinguish between luminal and non-luminal breast cancers at early stages before surgery based on pretherapeutic US images and biopsy H&E-stained WSIs, providing a tool to facilitate treatment decision making in early-stage breast cancers. Natural Science Foundation of Guangdong Province (No. 2023A1515011564), and National Natural Science Foundation of China (No. 91959127; No. 81971631).

The tumor suppressor CREBBP and the oncogene MYCN cooperate to induce malignant brain tumors in mice

The tumor suppressor and chromatin modifier cAMP response element-binding protein binding protein (CREBBP) and v-myc avian myelocytomatosis viral oncogene neuroblastoma derived homolog (MYCN), a member of the MYC oncogene family, are critically involved in brain development. Both genes are frequently mutated in the same tumor entities, including high-grade glioma and medulloblastoma. Therefore, we hypothesized that alterations in both genes cooperate to induce brain tumor formation. For further investigation, hGFAP-cre::CrebbpFl/Fl::lsl-MYCN mice were generated, which combine Crebbp deletion with overexpression of MYCN in neural stem cells (NSCs). Within eight months, these animals developed aggressive forebrain tumors. The first tumors were detectable in the olfactory bulbs of seven-day-old mice. This location raises the possibility that presumptive founder cells are derived from the ventricular-subventricular zone (V-SVZ). To examine the cellular biology of these tumors, single-cell RNA sequencing was performed, which revealed high intratumoral heterogeneity. Data comparison with reference CNS cell types indicated the highest similarity of tumor cells with transit-amplifying NSCs or activated NSCs of the V-SVZ. Consequently, we analyzed V-SVZ NSCs of our mouse model aiming to confirm that the tumors originate from this stem cell niche. Mutant V-SVZ NSCs showed significantly increased cell viability and proliferation as well as reduced glial and neural differentiation in vitro compared to control cells. In summary, we demonstrate the oncogenic potential of a combined loss of function of CREBBP and overexpression of MYCN in this cell population. hGFAP-cre::CrebbpFl/Fl::lsl-MYCN mice thus provide a valuable tool to study tumor-driving mechanisms in a key neural stem/ progenitor cell niche.

Effects of Heterogeneity on Cancer: A Game Theory Perspective

In this study, we explore interactions between cancer cells by using the hawk–dove game. We analyze the heterogeneity of tumors by considering games with populations composed of 2 or 3 types of cell. We determine what strategies are evolutionarily stable in the 2-type and 3-type population games and what the corresponding expected payoffs are. Our results show that the payoff of the best-off cell in the 2-type population game is higher than that of the best-off cell in the 3-type population game. When these mathematical findings are transferred to the field of oncology they suggest that a tumor with low intratumor heterogeneity pursues a more aggressive course than one with high intratumor heterogeneity. Some histological and genomic data on clear cell renal cell carcinomas is consistent with these results. We underline the importance of identifying intratumor heterogeneity in routine practice and suggest that therapeutic strategies that preserve heterogeneity may be promising as they may slow down cancer growth.

EPEN-12. COMPARING CELLULAR HIERARCHIES AND SPATIAL ARCHITECTURE ACROSS MOLECULAR SUBGROUPS OF SUPRATENTORIAL EPENDYMOMAS

Abstract Ependymomas (EPN) are among the most common and fatal pediatric brain tumors with currently no targeted therapies available. Molecular studies including genome-wide DNA methylation profiling have shown that supratentorial ependymoma is subdivided into subgroups characterized by distinct fusion proteins: ZFTA-RELA fusion-positive (ZFTA-RELA), YAP1 fusion-positive (ST-YAP1), and four alternative ZFTA fusion-positive (ZFTA-Cluster 1, ZFTA-Cluster 2, ZFTA-Cluster 3, and ZFTA-Cluster 4). ZFTA-RELA tumors have been identified as the most common and aggressive molecular group with high intra-tumoral heterogeneity. However, limited sample size and interpatient variability in previous studies highlight a critical need for comprehensive profiling of supratentorial ependymomas across a larger patient cohort. In particular, how ZFTA-RELA compares to other subgroups of supratentorial ependymoma is unknown. Here, we collected and profiled 44 supratentorial patient tumors encompassing ZFTA-RELA (n=21), ZFTA-Cluster 1 (n=6), ZFTA-Cluster 2 (n=4), ZFTA-Cluster 3 (n=7), ZFTA-Cluster 4 (n=2), and ST-YAP1 (n=4). We profiled all tumors by single cell/single nucleus RNA-sequencing, and further characterized ZFTA-RELA by single cell spatial transcriptomics. Within each patient’s tumor we find multiple normal and aberrant differentiation trajectories from neural progenitor-like cells towards glial-like, mesenchymal-like and ependymal-like cells. Using sequencing-based lineage tracing, we experimentally validate these computationally inferred cellular hierarchies in vitro. Notably, compared to ZFTA-RELA ependymomas we observed predominantly higher proportion of early neural progenitor-like cells in ZFTA-Cluster 2 and ZFTA-Cluster 3, while in ST-YAP1 greater representation of mature ependymal-like cells. These findings highlight differences in developmental and aberrant differentiation trajectories driven by combination of oncogenic fusions and cell of origin in supratentorial ependymoma subgroups. In addition, neural progenitor-like cells most highly express ZFTA-fusions and are spatially distinct from other cell states, highlighting oncogenic fusions driving tumorigenesis organized in spatial niches. Together, this study reveals previously unknown intra- and intertumoral heterogeneity across subgroups of supratentorial ependymoma, as well as distinct cellular and spatial architecture.

BIOL-10. MOUSE MODELING OF PEDIATRIC MYCN GLIOMA REVEALS INTRATUMORAL HETEROGENEITY INCLUDING NEURONAL AND OLIGODENDROGLIAL LINEAGE SIGNATURES

Abstract Pediatric glioma of the subclass MYCN are highly aggressive tumors with a median age at diagnosis of 8 years and a median survival of only 14 months. Molecularly, they frequently carry MYCN amplifications, TP53 mutations or both of these alterations. Due to their rarity, such tumors have only recently been identified as a distinct entity, and both biological as well as clinical details have not been fully addressed. In order to further investigate the biology and treatment options of these tumors, preclinical models are urgently needed. Therefore, we generated a genetically engineered model by breeding hGFAP-cre::Trp53Fl/Fl::lsl-MYCN mice, which carry a loss of Trp53 and show aberrant MYCN expression in neural precursors of the central nervous system. All such mice developed aggressive forebrain tumors early in lifetime that mimic their human counterparts regarding histology, DNA methylation, and gene expression. Single cell RNA sequencing revealed a high intratumoral heterogeneity with neuronal and oligodendroglial lineage signatures. To find new treatment options, we performed in vitro high-throughput drug screening on both, mouse and human tumor cells. Among 640 investigated anticancer compounds, Doxorubicin, Irinotecan, and Etoposide effectively suppressed the growth of tumor cells. As FDA-approved chemotherapeutics, these compounds are used for the treatment of other pediatric brain tumors. Repurposing them for the treatment of MYCN-driven high grade glioma could achieve a rapid transition to glioma treatment. We believe that our model will pave the way to improved treatment strategies for patients with these highly aggressive tumors.

DIPG-11. DISSECTING THE DIFFUSE MIDLINE GLIOMA TUMOR MICROENVIRONMENT COMMUNICATIONS USING MULTI-OMICS APPROACHES

Abstract Diffuse midline glioma (DMG) represents a highly aggressive pediatric brain tumor with no chance of survival. Treatment options are urgently needed, and efforts have been put towards developing T-cell immunotherapy. However, this should consider DMG’s unique tumor microenvironment (TME), both in terms of anatomical location and developmental stage. To gain knowledge on this unique tumor niche, we implemented an immuno-competent, syngeneic mouse model based on in-utero electroporation (IUE) of the key driver mutations to induce tumor formation at an embryonic stage. We characterized DMG’s spatial and molecular landscape by combining snRNAseq, spatial transcriptomics, and state-of-the-art imaging. We observed a high intratumoral heterogeneity with spatially restricted cell populations. Importantly, their molecular profiles correlate with patient data, illustrating the clinical relevance of the IUE model. Analysis of cell-cell communications in the TME revealed novel signaling between DMG and immunosuppressive myeloid cells that could impact T-cell response. Using an advanced patient-derived DMG organoid co-culture model with macrophages and engineered T-cells, we are currently investigating the role of this pathway on T-cells mediated killing of DMG organoids. Thus, this project fills a critical need to delineate and overcome the immune-suppressive environment in DMG and potentiate T cell targeting.

Intratumoral heterogeneity as a predictive biomarker in immunotherapy for advanced head and neck squamous cell carcinoma.

e18028 Background: Intratumoral heterogeneity (ITH) is associated with the poor prognosis in a variety of solid tumors, because of the therapeutic failure and drug resistance. However, we still lack understanding of the relationship between ITH and immunotherapy for advanced head and neck squamous cell carcinoma (HNSCC). Methods: We used the Bioconductor R package Maftools to analyze somatic variants of HNSCC from MSK-IMPACT-1611 study cohort (PMID: 30643254) with data from 68 patients. Then, a mutant-allele tumor heterogeneity (MATH) algorithm was used to measure ITH score. Finally, we analyzed the difference in the genomes, and the association with tumor mutational burden (TMB) between the ITH score high (ITH-H) and ITH score low (ITH-L) groups, and the effect of ITH on the prognosis of immunotherapy. Results: There was no significant difference in the high frequency mutated genes between the high ITH and low ITH groups. The genes with the highest mutation frequency in ITH-H group were TP53 (90.48%), TERT (47.62%), CDKN2A (38.1%), FAT1 (33.33%), NOTCH1 (28.57%), KMT2D (19.05%), APC (14.29%), and ASXL1 (14.29%), while the genes with the highest mutation frequency in ITH-L group were TP53 (44.68%), PIK3CA (38.3%), TRET (36.17%), FAT1 (21.28%), NOTCH1 (19.15%), KMT2C (17.02%), KMT2D (17.02%), and CDKN2A (14.89%). The median overall survival (mOS) of ITH-L group was significantly higher than that of ITH-H group (not reach versus 10 months, hazard ratio (HR) = 2.15, 95%CI 1.05−4.4, P = 0.032). There was no significant difference in median TMB levels between ITH-L and ITH-H groups (P = 0.55). The TMB high (TMB-H) patients had no significantly better mOS than the TMB low (TMB-L) group (HR = 0.54, 95%CI 0.25−1.16, P = 0.11). A pooled analysis of ITH and TMB show that the ITH-L+TMB-H group had significantly better improvement mOS than that in the ITH-L+TMB-L, ITH-H+TMB-H, and ITH-H+TMB-L group (HR = 0.71, 95%CI 0.13−1.35, P = 0.048). Conclusions: ITH may serve as a positive predictor of immunotherapy in patients with advanced HNSCC and is independent of TMB.

Evolution of genomic profiles in primary and recurrent brain gliomas.

e14042 Background: Diffuse gliomas are highly heterogenous tumors with variable biological behaviors, including high intratumor and extratumor heterogeneity, and with occurrence of recurrent lesions in majority of patients. During disease progression, gliomas undergo cellular and genomic evolution with newly acquired genetic properties. However, the mechanism of this complicated process associated with treatment failure is poorly understood. In this retrospective study, we performed cytogenomic analyses of primary and recurrent tumors in five patients with diffuse glioma who underwent surgical resections or biopsies of multiple recurrences. Methods: One primary and at least two recurrent freshly resected tumor tissues from each patient were analyzed using combination of cytogenomic methods. To assess specific and random copy number alterations (CNAs), I-FISH (Abbott Molecular, MetaSystems), array CGH/SNP (Agilent) and MLPA (MRC Holland) were performed. Methylation of MGMT promoter was investigated by methylation-specific MLPA (MRC Holland) and mutations of 67 genes associated with solid tumors were assessed by target NGS (Archer VariantPlex Solid Tumor, Invitae). Results: All five patients (three men and two women) experienced recurrence with newly acquired genetic or epigenetic changes. We observed a higher frequency of CNAs in recurrent gliomas than in primary tumors. Moreover, several aberrations were not detected in recurrence despite being found in earlier samples. As a primary event we proved mutation R132H of IDH1 gene. In addition, we detected methylation of the MGMT promoter, CDKN2A/B homozygous deletion, and RB1 deletion as later events that were probably associated with higher tumor grades. Besides the typical genomic changes, we detected many aberrations with unknown or unclear prognostic relevance (e.g. inframe deletion in TP53, p.Met243_Asn247del, etc.). The progression to a higher grade of glioma occurred in four patients. Conclusions: The evolutional patterns in glioma depend on clonal selection caused by CNAs, mutations, genetic drift, intratumor heterogeneity and/or the patient’s treatment. Recurrence may arise from one major tumor clone or from one or more subclones within the primary tumor through. Integrated cytogenomic analyses of genetic/epigenetic profiles of primary and all recurrent tissues can contribute to a better understanding of mechanisms responsible for these processes and to identification of new alterations related to gliomas progression and/or resistance to treatment. These biomarkers could subsequently serve as new therapeutic targets for personalized treatment.

Metabolomic differentiation of tumor core versus edge in glioma.

Gliomas exhibit high intratumor and interpatient heterogeneity. Recently, it has been shown that the microenvironment and phenotype differ significantly between the glioma core (inner) and edge (infiltrating) regions. This proof-of-concept study differentiates metabolic signatures associated with these regions, with the potential for prognosis and targeted therapy that could improve surgical outcomes. Paired glioma core and infiltrating edge samples were obtained from 27 patients after craniotomy. Liquid-liquid metabolite extraction was performed on the samples and metabolomic data were obtained via 2D liquid chromatography-mass spectrometry/mass spectrometry. To gauge the potential of metabolomics to identify clinically relevant predictors of survival from tumor core versus edge tissues, a boosted generalized linear machine learning model was used to predict metabolomic profiles associated with O6-methylguanine DNA methyltransferase (MGMT) promoter methylation. A panel of 66 (of 168) metabolites was found to significantly differ between glioma core and edge regions (p ≤ 0.05). Top metabolites with significantly different relative abundances included DL-alanine, creatine, cystathionine, nicotinamide, and D-pantothenic acid. Significant metabolic pathways identified by quantitative enrichment analysis included glycerophospholipid metabolism; butanoate metabolism; cysteine and methionine metabolism; glycine, serine, alanine, and threonine metabolism; purine metabolism; nicotinate and nicotinamide metabolism; and pantothenate and coenzyme A biosynthesis. The machine learning model using 4 key metabolites each within core and edge tissue specimens predicted MGMT promoter methylation status, with AUROCEdge = 0.960 and AUROCCore = 0.941. Top metabolites associated with MGMT status in the core samples included hydroxyhexanoycarnitine, spermine, succinic anhydride, and pantothenic acid, and in the edge samples metabolites included 5-cytidine monophosphate, pantothenic acid, itaconic acid, and uridine. Key metabolic differences are identified between core and edge tissue in glioma and, furthermore, demonstrate the potential for machine learning to provide insight into potential prognostic and therapeutic targets.