Subclonal Diversity Research Articles

Estimation of cancer cell fractions and clone trees from multi-region sequencing of tumors.

Multi-region sequencing of solid tumors can improve our understanding of intratumor subclonal diversity and the evolutionary history of mutational events. Due to uncertainty in clonal composition and the multitude of possible ancestral relationships between clones, elucidating the most probable relationships from bulk tumor sequencing poses statistical and computational challenges. We developed a Bayesian hierarchical model called PICTograph to model uncertainty in assigning mutations to subclones, to enable posterior distributions of cancer cell fractions (CCFs) and to visualize the most probable ancestral relationships between subclones. Compared with available methods, PICTograph provided more consistent and accurate estimates of CCFs and improved tree inference over a range of simulated clonal diversity. Application of PICTograph to multi-region whole-exome sequencing of tumors from individuals with pancreatic cancer precursor lesions confirmed known early-occurring mutations and indicated substantial molecular diversity, including 6-12 distinct subclones and intra-sample mixing of subclones. Using ensemble-based visualizations, we highlight highly probable evolutionary relationships recovered in multiple models. PICTograph provides a useful approximation to evolutionary inference from cross-sectional multi-region sequencing, particularly for complex cases. https://github.com/KarchinLab/pictograph. The data underlying this article will be shared on reasonable request to the corresponding author. Supplementary data are available at Bioinformatics online.

Emerging Biomarkers for Diagnosis, Prevention and Treatment of Brain Metastases-From Biology to Clinical Utility.

Primary malignancies of the lung, skin (melanoma), and breast have higher propensity for metastatic spread to the brain. Advances in molecular tumour profiling have aided the development of targeted therapies, stereotactic radiotherapy, and immunotherapy, which have led to some improvement in patient outcomes; however, the overall prognosis remains poor. Continued research to identify new prognostic and predictive biomarkers is necessary to further impact patient outcomes, as this will enable better risk stratification at the point of primary cancer diagnosis, earlier detection of metastatic deposits (for example, through surveillance), and more effective systemic treatments. Brain metastases exhibit considerable inter- and intratumoural heterogeneity—apart from distinct histology, treatment history and other clinical factors, the metastatic brain tumour microenvironment is incredibly variable both in terms of subclonal diversity and cellular composition. This review discusses emerging biomarkers; specifically, the biological context and potential clinical utility of tumour tissue biomarkers, circulating tumour cells, extracellular vesicles, and circulating tumour DNA.

KRAS mutation methylation clonality in early-stage pancreatic cancer.

614 Background: KRAS is mutated in 90% of pancreatic cancers making it a seemingly ideal target for treatment and yet, with the exception of the rare G12C mutation, KRAS is undruggable for the vast majority of pancreatic patients. Herein, we characterize, in silico, KRAS methylation-derived mutation clonal and subclonal diversity in PDAC and examine their impact upon clinical outcomes. Methods: We developed a mutation methylation (MM) clonality workflow for gene mutation assignment as clonal versus subclonal and applied it genome-wide to TCGA data. For comparison, we used the cancer cell fraction (CCF) clonality prediction. We examined clinical outcomes by comparing in months (mos), Kaplan-Meier estimated overall survival (OS) using a log-rank test and a cox model for testing several features. We performed differential gene expression, differential gene correlation, and gene set enrichment analyses (GSEA) between KRAS MM clonal versus subclonal early stage pancreatic patients. Results: Using 104 TCGA early stage pancreatic cancer patient tumors from TCGA with mutation, methylation and clinical outcomes data, we defined KRAS MM clonality (n = 70 clonal, n = 34 subclonal) and CCF clonality (n = 74 clonal, n = 28 subclonal) tumors. Clonality assignment between methods was 53% clonal and 17% subclonal concordant, and 19% discordant among samples. KRAS MM clonality was associated with significantly (p = 0.046) shorter OS (median OS = 11.5 mos) as compared to the KRAS subclonal group (median OS = 15 mos). By comparison, KRAS CCF clonal and subclonal patient groups did not differ in their OS. When RNA-Seq derived subtypes for pancreatic cancer were included in a model with our KRAS MM clonality marker, only our marker remained as significantly associated with OS. Median KRAS gene expression was significantly (p = 0.01) higher in the KRAS MM clonal versus subclonal group. A GSEA showed enrichment of MYC targets in the KRAS clonal group. We identified 72, mostly protein coding genes residing on chromosomes 5q, 7p and 8p that correlated with KRAS gene expression only in the subclonal group. Conclusions: Our analyses shows a potential clonality dissection of the established 90% KRAS mutation rate in pancreatic cancer, which based on our MM workflow, may be dissected into 61% KRAS clonal and 30% KRAS subclonal. By assigning clonality based on another DNA data type using CCF, we obtain 65% KRAS clonal and 25% KRAS subclonal. Thus, regardless of which DNA-based workflow, overall, the KRAS clonality rates are similar. There is a notable difference however in patient-level assignment of KRAS clonality as only our workflow showed poor OS associated with KRAS clonality. The introduction of a methylation-based mutation clonality marker could prove invaluable when used in combination with methylation-based circulating tumor DNA assays for patient and treatment selection, and clinical trial monitoring of tumor responses.

Elucidation of the pathogenesis and treatment of acute myeloid leukemia in animal models

Acute myeloid leukemia (AML) is a heterogeneous cell population comprising genetically diverse sub-clones with significant differences in properties that vary from one patient to another. Since AML properties are similar to those of hematopoietic stem and myeloid cells, bone marrow as an organ responsible for the survival of AML-initiating cells has been proposed to be able to cause relapse following chemotherapy. Therefore, establishing in vivo experimental systems is critical for understanding the properties of AML cells and developing therapeutic strategies. In this review, the history, advantages, and disadvantages of mouse leukemia models wherein mouse cells are transformed by oncogenic events, including xenograft mice in which human AML cells are transplanted into immunodeficient mice, were introduced. Following which I described the development of chimeric antigen receptor cell therapy using human cytokines expressing the AML xenograft mice.

A rapidly reversible mutation generates subclonal genetic diversity and unstable drug resistance

Most genetic changes have negligible reversion rates. As most mutations that confer resistance to an adverse condition (e.g., drug treatment) also confer a growth defect in its absence, it is challenging for cells to genetically adapt to transient environmental changes. Here, we identify a set of rapidly reversible drug-resistance mutations in Schizosaccharomyces pombe that are caused by microhomology-mediated tandem duplication (MTD) and reversion back to the wild-type sequence. Using 10,000× coverage whole-genome sequencing, we identify nearly 6,000 subclonal MTDs in a single clonal population and determine, using machine learning, how MTD frequency is encoded in the genome. We find that sequences with the highest-predicted MTD rates tend to generate insertions that maintain the correct reading frame, suggesting that MTD formation has shaped the evolution of coding sequences. Our study reveals a common mechanism of reversible genetic variation that is beneficial for adaptation to environmental fluctuations and facilitates evolutionary divergence.

The DNA methylation landscape of multiple myeloma shows extensive inter- and intrapatient heterogeneity that fuels transcriptomic variability

BackgroundCancer evolution depends on epigenetic and genetic diversity. Historically, in multiple myeloma (MM), subclonal diversity and tumor evolution have been investigated mostly from a genetic perspective.MethodsHere, we performed an analysis of 42 MM samples from 21 patients by using enhanced reduced representation bisulfite sequencing (eRRBS). We combined several metrics of epigenetic heterogeneity to analyze DNA methylation heterogeneity in MM patients.ResultsWe show that MM is characterized by the continuous accumulation of stochastic methylation at the promoters of development-related genes. High combinatorial entropy change is associated with poor outcomes in our pilot study and depends predominantly on partially methylated domains (PMDs). These PMDs, which represent the major source of inter- and intrapatient DNA methylation heterogeneity in MM, are linked to other key epigenetic aberrations, such as CpG island (CGI)/transcription start site (TSS) hypermethylation and H3K27me3 redistribution as well as 3D organization alterations. In addition, transcriptome analysis revealed that intratumor methylation heterogeneity was associated with low-level expression and high variability.ConclusionsWe propose that disrupted DNA methylation in MM is responsible for high epigenetic and transcriptomic instability allowing tumor cells to adapt to environmental changes by tapping into a pool of evolutionary trajectories.

Paracrine Behaviors Arbitrate Parasite-Like Interactions Between Tumor Subclones.

Explaining the emergence and maintenance of intratumor heterogeneity is an important question in cancer biology. Tumor cells can generate considerable subclonal diversity, which influences tumor growth rate, treatment resistance, and metastasis, yet we know remarkably little about how cells from different subclones interact. Here, we confronted two murine mammary cancer cell lines to determine both the nature and mechanisms of subclonal cellular interactions in vitro. Surprisingly, we found that, compared to monoculture, growth of the “winner” was enhanced by the presence of the “loser” cell line, whereas growth of the latter was reduced. Mathematical modeling and laboratory assays indicated that these interactions are mediated by the production of paracrine metabolites resulting in the winner subclone effectively “farming” the loser. Our findings add a new level of complexity to the mechanisms underlying subclonal growth dynamics.

Germinal Center Program De-Synchronization and Intra-Patient Heterogeneity in Follicular Lymphoma B-Cells Revealed By Integrative Single-Cell Analysis

IntroductionFollicular lymphoma (FL) is an indolent B-cell neoplasm with histological, phenotypical and transcriptional features of germinal center (GC) B-cells, suggesting that FL B-cells are “frozen” in a GC B-cell state. In normal GC B-cells, affinity maturation relies on iterative cycles of somatic hypermutation and proliferation followed by affinity-based selection. Thus GC B-cells are functionally heterogeneous. Although FL sub-clonal genetic diversity is well documented, tumor cell functional heterogeneity in FL has not been studied in detail. Bulk-level transcriptomic analyses fail to capture cellular heterogeneity and are thus insufficient to compare FL B-cells to the heterogeneous GC B-cell subset. Here, we used integrative single-cell analysis of phenotype, gene expression and IGH sequence to assess and compare functional heterogeneity in FL B-cells and GC B-cells.MethodsNormal B-cells were sorted from spleens or tonsil of healthy human donors (n=4), and FL B-cells from diagnostic lymph node biopsies (n=5), all available as frozen viable cell suspensions. Single normal or malignant B-cells were sorted by flow cytometry for parallel analysis of phenotype (9-color panel), gene expression (91-gene panel by microfluidic RT-qPCR), and IGH sequence (RT-PCR followed by Sanger sequencing; FL B-cells only). Custom and existing bioinformatics analysis pipelines were combined for quality control and cell filtering, dimensionality reduction (PCA, t-SNE), clustering, pseudo-time analysis, IGH sequence analysis and integrative data analysis.ResultsIn normal B-cells, single-cell gene expression profiling segregated major cell types (antibody-producing cells, memory B-cells and GC B-cells) and enabled the reconstruction of differentiation trajectories within cell types. Specifically, our data revealed a cyclic continuum of transitional GC B-cell states where synchronized expression of gene modules characterized human GC B-cell functional heterogeneity. Single-cell gene expression profiles defined FL B-cells as an independent cell type, distinct from GC B-cells. In particular, the expression of most characteristic GC-associated genes was de-synchronized in FL B-cells from all patients. Clustering of single FL B-cell 91-gene expression profiles revealed inter-patient and intra-patient heterogeneity, and defined common FL B-cell subsets found in most patients. Intra-patient heterogeneity was not linked to sub-clonal divergence as defined by IGH sequencing.ConclusionsUsing integrative single-cell analyses, we have shown that synchronized expression of characteristic gene modules drives GC B-cells through a cyclic continuum of transitional states. De-synchronization of that program in FL B-cells, and substantial intra-patient gene expression heterogeneity, suggest that FL B-cells are functionally plastic and not blocked to a GC B-cell state. DisclosuresNo relevant conflicts of interest to declare.

Progression to AML Is Predictable and Distinct from Age Related Clonal Hematopoiesis

Introduction: Despite marked improvements in our understanding of the biology and genetic basis of acute myeloid leukemia (AML) gained in the past 4 decades, the overall survival of AML has only improved slightly. The strategy for targeting AML once the disease is diagnosed remains grim due to subclonal genetic diversity and the stemness properties of relapse originating cell types. As studies in solid tumors have shown, strategies focused on targeting the cancer at earlier stages hold considerable promise, however no predictive tests are available for AML. The existence of clonally expanded pre-leukemic hematopoietic stem and progenitor cells (preL-HSPC) within the diagnostic blood sample of AML patients provides strong evidence that the disease must evolve over long time periods prior to diagnosis. However, with the discovery that age related clonal hematopoiesis (ARCH) is exceedingly common during normal aging; a refined search for the factors that distinguish ARCH from those rare individuals who will actually develop AML had to be taken.Methods: We hypothesized that the basis for clonal expansion in ARCH was different than in the clonal expansion leading to AML. To investigate genetic differences, we developed sensitive and accurate error correction sequencing method to study blood samples of healthy individuals from the EPIC cohort before AML diagnosis. These 96 pre-AML cases donated their blood on average 6 years before eventually being diagnosed with AML; a set of 420 matched controls were sequenced in the same way. Furthermore, 3.2 million electronic health records (EHR) from the Clalit database were inspected to identify non-genetic AML-risk prediction parameters.Results: The overall variant allele frequency (VAF) distribution of the identified mutations was significantly different between cases and controls. Specifically, a greater proportion of cases carried somatic mutations with intermediate-high VAF (70.8% versus 36.6% in the controls). The median number of all somatic mutations was significantly higher in the cases as compared with the controls (7 and 5 respectively) and was correlated with aging in both groups. Interestingly, the pre-AML group had a higher rate of mutation accumulation compared to controls, resulting in significantly greater numbers of mutations occurring after the age of 55. Recurrent AML mutations were more prevalent in cases compared to controls (Age>=55: OR, 7.571; 95% CI, 4.055-14.387). Specifically, highly recurrent splicing factors mutations (e.g. SRSF2 P95H, U2AF1 Q84P, and SF3B1 K700E) were found solely in the pre-AML group suggesting that specific mutations that rewire the splicing machinery inevitably leads to AML when acquired in HSPC. The VAF of recurrent AML mutations was significantly higher in the cases compared to controls. Furthermore we observed a higher proportion of individuals with more than one recurrent AML mutation in the pre-AML group compared to controls (OR, 14.316; 95% CI, 6.318-34.933). Incorporating these factors, we construct a computational model for predicting progression toward AML. The AML risk prediction model predicted progression on average 7 years before actual diagnosis (HR, 20.1, 95% CI 8.59-46.8). EHR data was found to be highly valuable for the identification of high-risk population. Compared to controls, AML patients (N=982) showed aberrant blood counts up to 6 months prior diagnosis (relative risk, 22.2).Conclusions: These results reveal that the clonal expansion occurring during normal aging is highly distinct from pre-AML clonal expansion, which has a highly predictable evolutionary trajectory. Our study provides clear proof of concept for the feasibility of early AML prediction based on somatic mutation patterns, yet, since ARCH is common, the real-world feasibility of applying our AML risk prediction model will require identifying higher-risk individuals through routine clinical testing. We have shown that information obtained from EHR is useful for that purpose. Future studies of independent population cohorts with access to serial viable blood and leukemia samples will allow for incorporation of information such as the specific identity of the mutated cells and clone expansion kinetics. Such studies will aid in developing better understanding of which individuals should be selected for future clinical trials to study the potential benefits of early interventions in this deadly disease. DisclosuresNo relevant conflicts of interest to declare.

Spatially Resolved Multi-Omics Deciphers Bidirectional Tumor-Host Interdependence in Glioblastoma

Glioblastomas are malignant tumors of the central nervous system hallmarked by subclonal diversity and dynamic adaptation amid developmental hierarchies. The source of the dynamic reorganization within the spatial context of these tumors remains elusive. Here, we characterized glioblastomas in-depth by spatially resolved transcriptomics, metabolomics and proteomics. By deciphering regional shared transcriptional programs across patients, we infer that glioblastomas are organized by spatial segregation of lineage states and adapt to inflammatory or metabolic stimuli reminiscent of reactive transformation in mature astrocytes. Integration of metabolic imaging and image mass cytometry uncovered locoregional tumor-host interdependence resulting in spatially exclusive adaptive transcriptional programs. Inferring copy-number alterations emphasizes a spatially cohesive organization of subclones associated with reactive transcriptional programs, confirming that environmental stress gives rise to selection pressure. A model of glioblastoma stem cells implanted into human and rodent neocortical tissue mimicking various environments confirmed that transcriptional states originate from dynamic adaptation to various environments. Funding: DHH is funded by the Else Kroner-Fresenius Foundation. The work is part of the MEPHISTO project (PI: DHH and DD), funded by BMBF (iGerman Ministry of Education and Research) (project number: 031L0260B). The Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) supports the work of AW (project number: SFB 824 C04). VR, KJ and UGH funded by BMBF (Bundes Ministerium fur Bildung und Forschung) (project number: FMT 13GW0230A), US was supported by the Fordergemeinschaft Kinderkrebszentrum Hamburg. We thank Dietmar Pfeifer for here helpful advice. We thank Biorender.com. We thank Stella Maria Carro for her support and the provision of her laboratory facilities and equipment. Declaration of Interest: No potential conflicts of interest were disclosed by the authors. Ethical Approval: The local ethics committee of the University of Freiburg approved the data evaluation, imaging procedures and experimental design (protocol 100020/09 and 472/15_160880).

Serial single-cell genomics reveals convergent subclonal evolution of resistance as early-stage breast cancer patients progress on endocrine plus CDK4/6 therapy.

Combining cyclin-dependent kinase (CDK) inhibitors with endocrine therapy improves outcomes for metastatic estrogen receptor positive (ER+) breast cancer patients but its value in earlier stage patients is unclear. We examined evolutionary trajectories of early-stage breast cancer tumors, using single cell RNA sequencing (scRNAseq) of serial biopsies from the FELINE clinical trial (#NCT02712723) of endocrine therapy (letrozole) alone or combined with the CDK inhibitor ribociclib. Despite differences in subclonal diversity evolution across patients and treatments, common resistance phenotypes emerged. Resistant tumors treated with combination therapy showed accelerated loss of estrogen signaling with convergent up-regulation of JNK signaling through growth factor receptors. In contrast, cancer cells maintaining estrogen signaling during mono- or combination therapy showed potentiation of CDK4/6 activation and ERK upregulation through ERBB4 signaling. These results indicate that combination therapy in early-stage ER+ breast cancer leads to emergence of resistance through a shift from estrogen to alternative growth signal-mediated proliferation.

Highlights from Recent Cancer Literature

Highlights from Recent Cancer Literature

Evolutionary Trajectories and Genomic Divergence in Localized Breast Cancers after Ipsilateral Breast Tumor Recurrence.

Simple SummaryIpsilateral breast tumor relapse (IBTR) occurs in 5–10% of localized breast cancers (BCs) within 10 years of incidence, despite proper treatment of the primary lesion. However, the clinical consequences of evolutionary trajectories of BC cells and their impact on IBTR remain poorly understood. Here, we conducted a longitudinal genomic analysis of 10 matched localized BC patients with IBTR. Overall, we identified the differences in homologous recombination deficiency, chromosomal instability, and somatic mutation drivers between primary and relapsed lesions. Our analyses highlighted three clonal architectures that shape by distinct mutagenic processes and subclonal diversification during relapse progression. Finally, this study provided a framework, which integrated actionable biomarkers with clonal architectures, towards improvement of future treatment decisions.The evolutionary trajectories that drive clinical and therapeutic consequences in localized breast cancers (BCs) with ipsilateral breast tumor relapse (IBTR) remain largely unknown. Analyses of longitudinal paired whole-exome sequencing data from 10 localized BC patients with IBTR reveal that, compared to primary breast tumors, homologous recombination (HR) deficiency, inactivation of the HR pathway, chromosomal instability, and somatic driver mutations are more frequent. Furthermore, three major models of evolution in IBTR are summarized, through which relative contributions of mutational signatures shift, and the subclonal diversity expansions are shown. Optimal treatment regimens are suggested by the clinically relevant molecular features, such as HR deficiency (20%) or specific alterations (30%) with sensitivity to available FDA-approved drugs. Finally, a rationale for the development of the therapeutic management framework is provided. This study sheds light on the complicated evolution patterns in IBTR and has significant clinical implications for future improvement of treatment decisions.

Breast tumours maintain a reservoir of subclonal diversity during expansion.

SummaryOur knowledge of copy number evolution during the expansion of primary breast tumors is limited1,2. To investigate this process, we developed a single cell, single-molecule DNA sequencing method and performed copy number analysis of 16,178 single cells from 8 triple-negative breast cancers (TNBCs) and 4 cell lines. Our data shows that breast tumors and cell lines are comprised of a large milieu of subclones (7–22) that are organized into a few (3–5) major superclones. Evolutionary analysis suggests that after clonal TP53 mutations, multiple LOH events and genome doubling, there was a period of transient genomic instability followed by ongoing copy number evolution during the primary tumor expansion. By subcloning single daughter cells in culture, we show that tumor cells re-diversify their genomes and do not retain isogenic properties. These data show that TNBCs continue to evolve chromosome aberrations and maintain a reservoir of subclonal diversity during primary tumor growth.

Bone marrow involvement and subclonal diversity impairs detection of mutated CXCR4 by diagnostic next-generation sequencing in Waldenström macroglobulinaemia.

CXCR4 mutations impact disease presentation and treatment outcomes in Waldenström macroglobulinaemia (WM). Non-uniform testing for CXCR4 mutations may account for discordant findings in WM clinical trials. We compared two approaches used in these trials for detection of the most common CXCR4 (S338X) variant: targeted next-generation sequencing (NGS) using unselected bone marrow (BM) samples, and combined allele-specific polymerase chain reaction (AS-PCR) and Sanger sequencing with unselected and CD19-selected BM samples. Our findings showed that targeted NGS frequently yielded false-negative results. Both CD19 selection and AS-PCR markedly improved detection of CXCR4S338X mutations. Sensitivity was adversely impacted by low BM involvement and CXCR4 mutation clonality.

Abstract PO-010: Elucidating intratumoral heterogeneity and predicting effective combination regimens in PDAC PDOs using scRNA-seq and snATAC-seq

Abstract Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related death in the United States. The overwhelming majority of patients (up to 85%) present with unresectable disease, and the 5-year survival rate stands at a dire 9%. Current chemotherapeutic regimens extend survival by only a few months, and the molecular underpinnings of treatment failure in advanced disease remain largely unknown. Recently, single-cell sequencing techniques have emerged as highly effective methods to uncover intratumoral heterogeneity across many cancer types, including PDAC. These studies, and our own unpublished data, have reiterated profound transcriptomic subclonal diversity in PDAC, which likely underlies the divergent therapeutic responses and eventual treatment failure. Using a patient-derived organoid (PDO) platform, we have characterized, on the single cell level, gene expression and regulation profiles to understand the heterogeneity and clinical utility of this model. We applied predictive RNA-seq-based algorithms (CMap) to our initial PDO profiling data to design “custom” PDO combinatorial regimens as a true precision medicine-based approach. We are in the process of validating these predictions in vitro; however, preliminary single-cell RNA (scRNA-seq) data shows treatment PDOs with CMap-predicted compounds induced dramatic transcriptomic shifts in subclones reflective of an efficacious treatment strategy. Additionally, we conducted single-nuclei ATAC sequencing (snATAC-seq) of PDOs pre- and post-treatment with CMap-predicted compounds. This data, coupled to our scRNA-seq, allowed for construction of chromatin regulatory maps, and determined alterations in chromatin accessibility and gene regulation induced by targeted combinatorial therapeutics. These findings provide meaningful insight to biological processes driven by treatment-induced chromatin accessibility, and its relationship to heterogeneous transcriptional architecture. Our data suggests that differential regulation across PDO subclones may drive responses to therapy and acquired resistance mechanisms observed in the clinic, which could be predicted and therapeutically targeted with scRNA sequencing. Thus, this work represents the importance of leveraging subclonal architecture in the preclinical space to ensure both treatment efficacy and the consideration of novel therapeutic combinations in PDAC patients. Citation Format: Paola A. Guerrero, Anirban Maitra, Maria E. Monberg, Jaewon J Lee, Vincent Bernard-Pagan, Senthil Muthuswamy. Elucidating intratumoral heterogeneity and predicting effective combination regimens in PDAC PDOs using scRNA-seq and snATAC-seq [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2020 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2020;80(22 Suppl):Abstract nr PO-010.

The Clonal Trajectories of SF3B1 Mutations in Myeloid Neoplasia

Molecular lesions have diagnostic and prognostic impacts in myeloid neoplasia (MN). The beau idéal is the presence of SF3B1MT in MDS with ringed sideroblasts. Although most of these patients have a classic phenotype, little is known about the factors diverting it during the clonal evolution. Clinical trajectories of patients with SF3B1MT might depend on clonal hierarchy, dynamics and subclonal diversity in relation to other lesions. Herein, we studied the molecular architecture of patients with SF3B1MT to determine whether clonality and rank might infer distinct MDS features. We collected molecular and clinical information of 3561 patients with MN from Cleveland Clinic and publicly shared data. Results of targeted deep sequencing of 176 MDS/ AML genes were included. We applied an in-house variant allele frequency (VAF) based bioanalytic method to resolve clonal hierarchy. VAF (adjusted for copy number and zygosity) was used to classify the mutations into dominant (if a cutoff of at least 5% difference between VAFs existed), secondary (any subsequent subclonal hit) and codominant hits (<5% difference of VAFs between 2 hits). SF3B1 MT were detected in 9% of patients (n=335): 140 were SF3B1DOM (42%), 121 were SF3B1SEC (36%) and 74 were codominant (SF3B1COD, 22%). Comparison of SF3B1MT VAFs among the 3 groups showed no significant difference (40 vs 39 vs 44%, P= .6). Due to lack of distinct partition, we delved into the implications of SF3B1COD on mutational stability, disease phenotypes, molecular associations and overall survival (OS) as compared to the other 2 configurations. Serially-collected samples (n=21) showed clear scenarios of dominant clones remaining stable or switching to secondary due to impending acquisition of other hits or secondary clones remaining as such but denoted an ambivalent framework of codominant clones with "sweeping" features. Cases with SF3B1COD had a trend of OS similar to that of SF3B1SEC (median OS: 18 vs 15.9 mo.) and as such a poorer outcome compared to that of SF3B1DOM (39.8 mo.; P= .02). Moreover, the mutational profile of SF3B1CODvsSF3B1DOM was partially similar to that of SF3B1SECvs.SF3B1DOM with significantly higher odd ratios for DNMT3A (P< .0001), ASXL1, FLT3, RUNX1, TET2 (P≤ .05 for each). No variant morphologic features were observed when comparing SF3B1COD to the other 2 groups. By virtue, we strictly compared only cases with SF3B1DOM and SF3B1SEC, further supported by our single cell analysis(n=5), to showan unambiguous distinction between these two statii. Mutational burden analysis revealed that SF3B1DOM patients had a fewer number of mutations than those with SF3B1SEC (median 1.0 vs 2.6). SF3B1DOM was mostly detected in patients with a normocellular bone marrow as compared to SF3B1SEC (46 vs 29%, P= .02) and had a lower proportion of multidysplastic myeloid cells (29 vs 53%, P= .01). On the other side, SF3B1SEC were often present in the context of bilineage dysplasia (47vs. 26%, P= .02). Focused morphologic analysis revealed that SF3B1DOM cases were significantly associated with a higher RS% (median 37 vs 3%, P= .002; frequency of RS ≥15%: 77 vs 44%, P= .003) with higher VAF% significantly correlating with higher RS%. Patients with SF3B1SEC had shorter OS than those with SF3B1DOM (15.9 vs 39.7 mo., P< .0001). Even by dichotomizing according to VAF, the median OS of cases with VAF>40% was significantly shorter than those with VAF<20% (14.4 vs 33.9 mo.; P= .001) and between 20% and 40% (39.3 mo.). When the VAF of SF3B1SEC was >40%, the OS was shortened compared to SF3B1DOM (31 vs 11.6 mo., P= .001) and similarly when it laid between 20% and 40% (49.7 vs 25.6 mo., P= .01) suggesting a strong impact of associated hits. In SF3B1SEC, univariate analyses showed significantly higher odds of hits in RUNX1 (43 vs 19%, P< .0001), TET2 (29 vs 11%, P= .0005), FLT3 (22 vs 11%, P= .02), DNMT3A (20 vs 7%, P= .004), ASXL1 (16 vs 5%, P= .06), BCOR/L1 (17 vs 5%, P= .005), IDH1/2 (11 vs 2%, P= .008), CBL (8 vs 1%, P= .009) and CEBPA (7 vs 2%, P= .04) compared to SF3B1DOM. Interestingly, cases with co-existing TET2 mutations had a marked decrease in OS in SF3B1SECvsSF3B1DOM(10.1 vs 96.1 mo., P= .02) suggesting that the mutational ranking in a disease triggered by SF3B1MT can be skewed by stronger hits. In sum, our study suggests that molecular ranking in the context of SF3B1 clonal configuration is a key factor diverting the clinical and phenotypic trajectories of patients with MN and SF3B1MT. Disclosures Carraway: BMS: Consultancy, Other: Research support, Speakers Bureau; Jazz: Consultancy, Speakers Bureau; Stemline: Consultancy, Speakers Bureau; Takeda: Other: Independent Advisory Committe (IRC); ASTEX: Other: Independent Advisory Committe (IRC); Abbvie: Other: Independent Advisory Committe (IRC); Novartis: Consultancy, Speakers Bureau. Sekeres:Pfizer: Consultancy; BMS: Consultancy; Takeda/Millenium: Consultancy. Maciejewski:Alexion, BMS: Speakers Bureau; Novartis, Roche: Consultancy, Honoraria.

Abstract PO-133: Breast tumors maintain a reservoir of subclonal diversity during primary expansion

Abstract Our knowledge of copy number evolution during the expansion of primary breast tumors is limited. To investigate this process, we developed a single-cell, single-molecule DNA sequencing method and performed copy number analysis of 13,808 single cells from six triple-negative breast cancers (TNBCs) and two breast cancer cell lines. Our data shows that breast tumors and cell lines are comprised of a large milieu (17-26) of subclones that are organized into a few (4-8) major superclones. Phylogenetic analysis and mathematical modeling show that copy number evolution was ongoing during the expansion of the primary tumor, after the initial punctuated burst of genomic instability. By expanding single daughter cells in culture, we show that cancer cells re-diversify their genomes and do not retain isogenic properties. These data elucidate modes of evolution and show that TNBCs maintain a vast reservoir of subclonal copy number diversity during the expansion of the primary tumor mass. Citation Format: Darlan Conterno Minussi, Michael Nicholson, Hanghui Ye, Alexander Davis, Kaile Wang, Emi Sei, Haowei Du, Mashiat Rabbani, Cheng Peng, Min Hu, Shanshan Bai, Thomas McDonald, Aislyn Schalck, Anna Casasent, Angelica Barrera, Hui Chen, Bora Lim, Banu Arun, Funda Meric-Bernstam, Franziska Michor, Nicholas Navin. Breast tumors maintain a reservoir of subclonal diversity during primary expansion [abstract]. In: Proceedings of the AACR Virtual Special Conference on Tumor Heterogeneity: From Single Cells to Clinical Impact; 2020 Sep 17-18. Philadelphia (PA): AACR; Cancer Res 2020;80(21 Suppl):Abstract nr PO-133.

Abstract PO-103: Characterizing epigenetic and genetic mechanisms of glioma drug resistance using simultaneous lineage tracing and single-cell transcriptomic analysis

Abstract Therapeutic pressure can result in tumor evolution and adaptation via genetic and epigenetic mechanisms, which may be transient or stable. Knowledge of how these events contribute to intratumoral heterogeneity and dynamic subpopulations will require single cell-resolution experimental approaches that prospectively label, longitudinally track, and functionally characterize adaptive subclonal populations. In glioblastoma, poor efficacy of therapies targeting receptor tyrosine kinases (RTKs) can be attributed to a variety of mechanisms including intrinsic genetic subclonal diversity and therapy-derived epigenetic transitions. To more fully elucidate the factors contributing to RTK inhibitor resistance in glioblastoma, we designed and employed an experimental approach that combines high diversity cell lineage barcoding with single cell transcriptomic analysis. When a PDGFRA-amplified glioblastoma stem cell (GSC) model was treated with the RTK/PDGFRA inhibitor dasatinib, we observed that a broad spectrum of subclones adopt a temporary, Notch-dependent persister phenotype that presumably sustains cells through the initial drug exposure phase. Continued exposure to dasatinib resulted in the outgrowth of rare subclones with preexisting or acquired genetic changes. Single-cell analyses revealed several different subclones bearing distinct copy number amplifications of the insulin receptor substrate-1 and -2 (IRS1 or IRS2) loci, which activate insulin and AKT signaling programs and have not yet been previously identified as genes involved in glioblastoma drug resistance. Analysis of the Cancer Genome Atlas (TCGA) database revealed that IRS copy number amplifications are evident in a subset of a range of primary tumors and are associated with poor prognosis in glioblastoma subtypes bearing PDGFRA-amplified tumors. Functional characterization confirms that IRS2-amplified subclones lose dependence on the Notch signaling that characterizes the persister phenotype, and instead become dependent on insulin receptor/AKT signaling. Overall, our method for combined lineage-tracing and scRNA-seq revealed interplay between complementary genetic and epigenetic mechanisms of resistance in a heterogeneous glioblastoma tumor model. Moreover, this method permitted identification of previously unappreciated genetic alterations that may underlie the inefficacy of targeted therapies in glioblastoma. Citation Format: Christine E. Eyler, Hironori Matsunaga, Volker Hovestadt, Samantha J. Vantine, Peter van Galen, Bradley E. Bernstein. Characterizing epigenetic and genetic mechanisms of glioma drug resistance using simultaneous lineage tracing and single-cell transcriptomic analysis [abstract]. In: Proceedings of the AACR Virtual Special Conference on Tumor Heterogeneity: From Single Cells to Clinical Impact; 2020 Sep 17-18. Philadelphia (PA): AACR; Cancer Res 2020;80(21 Suppl):Abstract nr PO-103.

Revealing the subtyping of non-small cell lung cancer based on genomic evolutionary patterns by multi-region sequencing.

Accurately classifying patients with non‐small cell lung cancer (NSCLC) from the perspective of tumor evolution has not been systematically studied to date. Here, we reconstructed phylogenetic relationships of somatic mutations in 100 early NSCLC patients (327 lesions) through reanalyzing the TRACERx data. Based on the genomic evolutionary patterns presented on the phylogenetic trees, we grouped NSCLC patients into three evolutionary subtypes. The phylogenetic trees among three subtypes exhibited distinct branching structures, with one subtype representing branched evolution and another reflecting the early accumulation of genomic variation. However, in the evolutionary pattern of the third subtype, some mutations experienced selective sweeps and were gradually replaced by multiple newly formed subclonal populations. The subtype patients with poor prognosis had higher intra‐tumor heterogeneity and subclonal diversity. We combined genomic heterogeneity with clinical phenotypes analysis and found that subclonal expansion results in the progression and deterioration of the tumor. The molecular mechanisms of subtype‐specific Early Driver Feature (EDF) genes differed across the evolutionary subtypes, reflecting the characteristics of the subtype itself. In summary, our study provided new insights on the stratification of NSCLC patients based on genomic evolution that can be valuable for us to understand the development of pulmonary tumor profoundly.