Cancer Evolution Research Articles

Abstract B043: Quantifying selection intensity and epistatic interactions among gene variants within angiosarcoma

Abstract Evolutionary processes underlie cancer development. Mutations occur and accumulate in somatic tissues, and some mutations result in increases in cell division or survival and fix in cell populations more often than expected by chance. Computational techniques that account for differences in mutation rate across the genome and between tissues have helped pinpoint which genes are contributing to tumor growth within a wide array of cancer types. However, rare cancers are often understudied due to lack of funding, availability of data, and applicable tools that are capable of incorporating data from different sources. Recent initiatives, such as the Angiosarcoma Project, have increased data availability for the study of angiosarcoma, a rare cancer with approximately 300 new cases a year. In this study, we incorporate whole-genome, whole-exome, and targeted sequencing data from different sources—accounting for study-specific differences in genome coverage—into an evolutionary analysis of angiosarcoma tumorigenesis. Prior to our study, potential genes with oncogenic activity for angiosarcoma have been primarily evaluated based on genetic variant prevalence among tumor samples, but substitution prevalence is dependent on intrinsic mutation rate and thus relative prevalence of variants may not correspond with relative tumorigenic potential. We calculate the intensity of selection, a measure of the magnitude of increase of cell division and survival, of individual variants driving angiosarcoma, and complement existing studies that have previously determined KDR and TP53 mutations to be a driver of angiosarcoma, and we also implicate novel putative drivers that have high calculated selection intensity, including BRAF V600E, IGDCC4 S1020A, and SETD2 Q1152E. Additionally, we evaluated epistatic interactions among genes and found that selection for TP53 is increased after mutation of other genes such as MUC16, FAT1, and NOTCH2. This study demonstrates the feasibility of applying computational models informed by evolutionary approaches to a rare disease as well as identifies gene variants that likely have a large impact on angiosarcoma tumorigenesis. Citation Format: Caralynn E. Hampson, Vincent L. Cannataro. Quantifying selection intensity and epistatic interactions among gene variants within angiosarcoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B043.

Abstract B041: Identification of m3C modification sites within mRNAs of pancreatic cancer cells by long-read direct RNA sequencing

Abstract Chemical modification is a precise and effective way to control the biogenesis, stability, and function of biological molecules, such as nucleic acids, proteins, and lipids. More than 170 different types of modifications have been discovered in coding and non-coding RNAs, constituting a diverse RNA modification landscape, called epitranscriptome. RNA modifications were mediated by three types of proteins: writers, erasers, and readers. Recently, RNA modification machinery has been dysregulated in several types of cancers and may be promising targets for cancer therapy. 3-methylcytidine (m3C), which is catalyzed by METTL2A, 2B, 6, and 8, is found at position 32 of cytoplasmic tRNAs in organisms ranging from yeast to humans, however, their functions and regulatory machinery in mRNA and non-coding RNA has not been clarified because there were no effective anti- m3C antibodies. Here we discovered that the expression of m3C writers was higher in pancreatic tumors than in noncancerous tissues. We further found that METTL2A expression is associated with a poor prognosis in pancreatic cancer patients. We also showed that METTL2A depletion led to a reduction in the growth of AsPC1 cells. We next attempted to identify the m3C sites in cellular transcripts regulated by METTL2A in AsPC1 cells using long-read direct RNA sequencing. We corrected the transcriptome using ESPRESSO and searched for the RNA modifications using Nanocompore. We then found that METTL2A targets the NYCYN motif in mRNAs and mitochondrial rRNAs. Moreover, we found that transcripts with METTL2A-mediated m3C modification were upregulated in AsPC-1 cells knocked down by METTL2A. Our study provides evidence for METTL2A-mediated down-regulation of target mRNAs via m3C modification in the survival of pancreatic tumor cells. Citation Format: Shuhei Mitsutomi, Anzu Sugawara, Kenji Takahashi, Yusuke Mizukami, Nobuyoshi Akimitsu, Kenzui Taniue. Identification of m3C modification sites within mRNAs of pancreatic cancer cells by long-read direct RNA sequencing [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B041.

Abstract B042: Analysis of predictive indicators of invasive breast cancer: Modeling survival rates

Abstract Objective: We aim to investigate indicators of risk for invasive breast cancer (IBC) and to predict and assess the impact of these risk indicators on survival rates in breast cancer (BC) patients who received various therapies. Background: It has been identified that several indicators including BC subtypes, age at diagnosis, history of breast disease, menopausal status, estrogen receptor (ER) status, progesterone receptor (PR) status, human epidermal growth receptor (HER2) status, tumor stage, tumor size, among others, can cause genomic alterations and driver mutations in BC, leading to invasive carcinogenesis. To assure improved BC outcomes, we investigate the influence of BC indicators on survival rates in patients receiving hormone therapy, chemotherapy and radiotherapy. Methods: This study uses the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset, which included 31 clinical indicators, m-RNA levels z-score for 331 genes, and mutation in 175 genes for 1904 BC patients. Both bivariate and multivariate associations were examined using logistic regression and Cox proportional hazard models. The multivariate logistic regression analysis is used to determine predictors of IBC. The Cox proportional hazard model was also used to evaluate the magnitude of the impacts of clinical and demographic variables on IBC, as well as the effects of therapy on survival. All analyses for this investigation were carried out using SAS. A p-value < 0.05 was considered statistically significant. Results: We examined 1,043 patients (with the median age 61±13.1 years) from a total of 1904 for this study: 78.8% of patients had breast invasive ductal carcinoma; 87.4% were HER2 negative; 22.8%, 60.7%, and 67.6% received chemotherapy, hormone therapy, and radiotherapy treatments, respectively; 56.9% died but only 34.6% died from the disease; the average overall survival months was 127.3±77.6 months; and the most common age group was "65+ years" (39.4%). Chemotherapy (p=0.0474), neoplasm histologic grade (p=0.0015), Nottingham prognostic index (p<.0001), radiation (p=0.0068), and BReast CAncer gene 1 (BRCA1) (p=0.0251) were all associated with IBC. Overall survival was also correlated to age at diagnosis (p=0.0002), neoplasm histologic grade (p<.0001), lymph nodes (p<.0001), mutation count (p<.0001), Nottingham prognostic index (p<.0001), radiotherapy (p=0005), tumor size (p<.0001), tumor stage (p<.0001), BRCA1 (p=0.250), PTEN (p<.0001), and TP53 (p=0.0248). The overall accuracy of the model for each IBC and survival status was 82.1% and 74.2%, respectively. Conclusion: Our analysis demonstrated the importance of predictive analysis using the logistic regression model and survival analysis using the Cox proportional hazard model in cancer research. Predictive and survival analysis results aided in predicting the invasiveness and survival rate of BC among the study participants. In general, predictive and survival analyses are more accurate and may be used to make more realistic treatment decisions and precision oncology. Citation Format: Olumide Arigbede, Gebre-Egziabhe Kiros. Analysis of predictive indicators of invasive breast cancer: Modeling survival rates [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B042.

Abstract A020: An early determination of patients eligibility for a bladder cancer immunotherapy using a data science approach

Abstract Advanced bladder cancer patients have limited treatment options and, in most cases, limited overall survival. The first clinical trial with adoptive T cell therapy (ACT) that uses bladder cancer patients’ autologous tumor-infiltrating T lymphocytes (TILs) has been recently opened at Moffitt. To help identify which patients could benefit from ACT-TIL we are developing a machine learning predictor that takes advantage of tumor evolution data, such as tumor stages and grades routinely collected in clinic. The first phase of ACT-TIL involves resection of the primary tumor and the growth of TILs followed by a rapid expansion protocol of the reactive T cells. It may take up to 4-5 weeks before bladder cancer patients can get their own expanded TILs. Our predictor aims to determine whether the tumor isolated from a patient will grow TILs. In this work, we present a machine learning predictor of expansion of tumor infiltrating lymphocytes (ML-PETIL) that uses retrospectively collected patient tumor data with either demographic, clinical, or biological sample-related features. The proposed method consists of a suite of algorithms to identify a robust set of predictive features (feature selection algorithms) that are used to discriminate the dataset into two classes-that with successful TIL expansion or with no TIL growth (classification algorithms). Our predictor includes metrics which are used to address data imbalance and handle small to medium-sized datasets that contain features of varying datatypes and magnitudes. The ML-PETIL protocol offers an early determination of patients who would be eligible for ACT-TIL therapy. Citation Format: Kayode D. Olumoyin, Ahmet M. Aydin, Brittany L Bunch, Shari Pilon-Thomas, Michael Poch, Katarzyna A. Rejniak. An early determination of patients eligibility for a bladder cancer immunotherapy using a data science approach [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A020.

Abstract A011: Modeling the coevolution of native and CAR T-cells in large B cell lymphoma reveals a potential biomarker for response to therapy

Abstract Application of CD19 targeted Chimeric Antigen Receptor (CAR) T-cell therapy to large B cell lymphoma (LBCL) has yielded unprecedented clinical outcomes in a cancer largely resistant to conventional chemotherapy. This is highlighted in the recent ZUMA-7 trial, where axicabtagene ciloleucel (axi-cel), a CAR T-cell therapy, yielded a complete response (CR) rate in 65% of patients, an improvement over the 32% CR rate when using standard chemotherapy. The ZUMA-7 trial highlights the potential of CAR T-cell therapy to treat traditionally resistant cancers, but work remains to be done to identify biomarkers predictive of response. Operating under the assumption that CAR T not only directly targets CD19+ tumor cells, but also enhances tumor kill by amplifying the native T-cell populations, we hypothesized that neoantigen burden (NAB) may serve as a useful biomarker to make predictions of response to CAR T. We explore this hypothesis by building upon traditional predator-prey mathematical models to simulate the evolution of LBCL under immune predation and treatment with CAR T-cell therapy. More specifically, in our model native T-cells prey upon tumor cells that carry their cognate neoantigen, and each T-cell has variable search rate proportional to the cognate neoantigen’s recognition potential, modeled using a type III functional response. Tumor cells may mutate during division, always gaining a neoantigen, but possibly also losing HLA function, paving the way for immune escape. In addition to modeling the direct killing of tumor by CAR T, we also explore the synergistic effect of CAR T on native T-cells by modeling increased amplification of, and killing by, native T-cells based on interactions with the CAR T-cell population. Preliminary simulations suggest a high NAB can indeed be predictive of CAR T response, although only if the tumor has not evolved an escape mechanism. Indeed, simulations suggest that high NAB can also be a signature of an immune escaped tumor, as neoantigens become neutral traits once the immune system is removed as a selective force. These results suggest that knowledge of a patients’ NAB and immune competency can be used to guide CAR T treatment decisions. Citation Format: Chandler D. Gatenbee, Mark Robertson-Tessi, Maximilian Strobl, Ryan O. Schenck, Bachisio Ziccheddu, Francesco Maura, Frederick Locke, Alexander R.A. Anderson. Modeling the coevolution of native and CAR T-cells in large B cell lymphoma reveals a potential biomarker for response to therapy [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A011.

Abstract IA016: Bayesian phylodynamics to quantify cancer growth across space and time

Abstract Phylogenetics has traditionally inferred shared genetic histories of species over macroevolutionary time. With the expansion of cancer multi-region and single-cell sequencing, phylogenetic approaches have found new uses mapping the evolutionary histories of tumors as well. Recently developed “phylodynamic” methods that integrate phylogenetic and population dynamic modeling offer further promise - uncovering cancer growth dynamics from the topologies and branch lengths of tumor trees. Here, I discuss two recent efforts in this area: 1) Heterogeneous growth rates across a tumor become encoded in the branching patterns of phylogenetic trees built from multi-region sequencing. We introduce a cancer-specific extension to a class of multi-state birth-death models to infer these growth rate differences across space. We apply our model (SDevo) to a sample of hepatocellular carcinomas and infer that these tumor peripheries grow 2-6 times faster than their interiors. 2) Tumors experience differential growth rates over time as they acquire new driver mutations and invade their surrounding tissues, and these growth dynamics similarly leave characteristic signatures in their phylogenies. We employ Bayesian skyline models to dissect tumor growth dynamics from a dataset of diverse cancers and reveal variation in cancer growth rates over time. As cancer sequencing continues to proliferate, phylodynamic approaches are poised to reveal new quantitative insights into cancer growth dynamics across time and space. Citation Format: Maya Lewinsohn, Trevor Bedford, Nicola F. Müller, Alison Feder. Bayesian phylodynamics to quantify cancer growth across space and time [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr IA016.

Abstract A025: Adaptive therapy: Leveraging evolutionary dynamics for colorectal cancer management in vivo

Abstract Therapeutic resistance to chemotherapy is frequently the proximate cause of cancer treatment failure and, ultimately, patient mortality. Evolutionary therapies have been developed to address this problem and potentially reduce toxicities. The most successful strategy to date, adaptive therapy (AT), is designed to use chemosensitive cells to out-compete and suppress the chemoresistant cells. However, detailed observations to elucidate the mechanisms of AT's success are lacking. Typically, AT involves lowering the dose of the chemotherapy, thereby reducing toxicity and preserving the immune system. We are evaluating AT for the first time in colorectal cancer, using syngeneic orthotopic xenograft models. This project marks several firsts: 1) sensitive and resistant clones will be tracked longitudinally, 2) AT will be tested with a targeted therapy (MRTX1133), 3) AT will be tested in an immune competent model, 4) multiple drugs will be used for AT, and 5) extinction therapy will be tested in mice. We will employ genetic barcodes and differential luciferase tagging to track the tumor cell subpopulations longitudinally. Currently, we are conducting an in-vivo pilot experiment to validate our protocol, emphasizing tumor engraftment efficacy and tracking cell subpopulations. Additionally, we are investigating the potential of Lactobacillus acidophilus to inhibit tumor growth, augment chemotherapy, and bolster immune response in colorectal cancer. Citation Format: Harley I. Richker, Gissel Marquez-Alcaraz, Carlo Maley. Adaptive therapy: Leveraging evolutionary dynamics for colorectal cancer management in vivo [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A025.

Abstract B008: The polyclonal path towards malignant transformation in familial adenomatous polyposis

Abstract The widely accepted paradigm that cancer initiation is a monoclonal event has not been definitively resolved. This is largely due to the limitations of experimental systems to study cancer initiation, a definitively transient step in tumorigenesis. Individuals with familial adenomatous polyposis (FAP) possess a germline APC mutation, leading to hundreds of colonic polyps in adolescence and an eventual 100% lifetime risk of colorectal cancer. As such, total colectomies are performed once polyp burden becomes unmanageable and provide the resected colons from FAP donors provide a source of histologically normal mucosa, benign and dysplastic polyps, and, on rare occasions, adenocarcinomas, and serve as an exemplary model disease to understand human tumorigenesis. Through whole genome sequencing of 123 independent samples from six patients, we reveal the polyclonal origins of tumorigenesis in FAP. We characterize and contrast the genomic landscape and relative timing of driver mutation acquisition across normal to dysplastic and invasive colon tissue and compare this with multi-region sequencing data from sporadic colorectal cancers (CRC). We find that only early sporadic CRC drivers are common across the more heterogeneous premalignant stages. Selection, as assessed based on the ratio of nonsynonymous to synonymous mutations (dN/dS), is higher in benign than dysplastic FAP lesions and in sporadic adenomas relative to carcinomas. Phylogenetic reconstruction and pairwise sample comparisons reveal extensive heterogeneity and subclonal sharing within a patient’s benign and dysplastic lesions. Using a mechanistic mathematical model, we infer polyclonality for 90-100% of benign and 53-83% of dysplastic samples. Based on these findings, we propose a conceptual framework for the polyclonal origin of intestinal malignancies where inconsequential mutations, arising during colon formation in utero, are present across the colon and are valuable markers to uncover polyclonal ancestry. Citation Format: Ryan O. Schenck, Aziz Khan, Aaron M. Horning, Edward D. Esplin, HTAN Consortia, William J. Greenleaf, James M. Ford, Michael P. Snyder, Christina Curtis. The polyclonal path towards malignant transformation in familial adenomatous polyposis [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B008.

Abstract PR008: Cell state transitions drive the evolution of disease progression in B-cell acute lymphoblastic leukemia

Abstract A mathematical framework is constructed to predict the risk of B-lymphoblastic leukemia (B-ALL) relapse post-induction chemotherapy. The framework is a Markov chain model that quantifies spontaneous cell state transitions between distinct immunophenotypic subpopulations defined by CD34 and CD38 expression relative to neutrophils. Cell states are analyzed via flow cytometry pre- and post-treatment, providing insight into the evolution of cell state transition rates during disease progression. Cancer stem cells (CSCs) are hypothesized to promote tumor progression through innate chemoresistance and self-renewal. Ostensible CSCs were first identified in acute myeloid leukemia and were found to have a CD34+/CD38- immunophenotype similar to hematopoietic stem cells. However, the isolation of CSCs from B-ALL has proved more difficult. B-ALL cells with stem cell-like properties have been reported with variable immunophenotype, perhaps due to temporal variation of CD34 and CD38 expression in this setting. We hypothesized that transitions between stem cell-like, hematogone-like, and naive B-cell-like leukemia subpopulations play a significant role in B-ALL disease progression. To test this hypothesis, we trained a Markov chain mathematical model using flow cytometry characterization of four B-ALL cell states with their normal counterpart appearing in parentheses: CD34+/CD38- (hematopoietic stem cells), CD34+/CD38+ (stage 1 hematogones), CD34-/CD38+ (stage 2 and 3 hematogones), and CD34-/CD38- (naïve B-cells). An iterative numerical search procedure was used to derive patient-specific Markov matrices, describing the stochastic cell state transitions. This flow cytometry evaluation was performed on a cohort of patient samples of peripheral blood (N=46) and bone marrow (N=63) with matched clinical features such as BCR::ABL1 status, comprehensive genomic profiling, minimal residual disease (MRD) post-induction chemotherapy, and 3-year relapse. Critical to our goal of quantifying the evolution of state transition rates, we also obtained bone marrow measurements for a cohort of normal/healthy individuals. Patients were divided into post-induction flow MRD positive (N=16), MRD negative (N=30), healthy (N=6) cohorts, as well as relapsed and non-relapsed cohorts to compare features of the transition matrices. Importantly, pre-treatment flow cytometry derived cell state distribution alone is not predictive of relapse or MRD. In contrast, pre-treatment Markov chain transition parameters are found to be clinically predictive of relapse and MRD. MRD correlates to high reciprocity (the product of incoming and outgoing transitions) of the stem cell state. This approach provides supporting evidence that cell state transitions drive B-ALL disease progression. There is an additional strong correlation between Markov transition parameters and both BCR::ABL1 and BCR::ABL1-like B-ALL classification. Finally, comparison of Markov parameters pre- and post-treatment quantifies the evolutionary selection pressures acting on transition rates induced by chemotherapy treatment. Citation Format: Curtis Gravenmier, Ling Zhang, Lynn Moscinski, Jeffrey West. Cell state transitions drive the evolution of disease progression in B-cell acute lymphoblastic leukemia [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr PR008.

Abstract PR007: Molecular subtyping of lung adenocarcinoma premalignant lesions identifies features associated with aggressive disease

Abstract Lung cancer is the leading cause of cancer-related death. The five year survival rate had remained stagnant for decades, but more effective screening protocols using low-dose computed tomographic screening for high-risk individuals, along with progress in treatment for advanced-stage lung cancer with targeted therapy and immunotherapy, have helped decrease lung cancer mortality. However, improved approaches to lung cancer prevention and early detection are still urgently needed. Adenocarcinoma (LUAD) is the most common form of lung cancer. Following surgical resection of pre-invasive adenocarcinoma in situ, the five-year survival rate approaches 100%. This emphasizes the need for improved understanding of characteristics of aggressive premalignant lesions that would benefit from early interventions. In contrast to the well-characterized histopathologic sequence of lung squamous cell carcinoma, atypical adenomatous hyperplasia (AAH) and adenocarcinoma in situ (AIS) are the only known precursors in the sequence of LUAD pathogenesis. The features of AAH and AIS that lead to invasive LUAD are poorly characterized. The relationship of AIS and AAH to the histologically diverse subtypes of LUAD, ranging from the typically slow growing lepidic to aggressive solid subtypes, are also unclear. We hypothesized that transcriptomic changes in subsets of premalignant lesions are linked to distinct genomic and clinicopathologic features of malignant disease. To investigate this hypothesis, we performed exome sequencing and bulk RNA sequencing of laser capture microdissected tissue from tumor margins that included premalignant lesions, invasive tumor, and adjacent normal tissues. With our analysis, we discovered de novo subtypes of LUAD premalignant lesions based on gene expression. One of these subtypes had gene expression alterations similar to aggressive invasive LUAD solid tumors. This molecular adenomatous premalignant lesion subtype was also associated with an increased accumulation of cancer driver mutations. This subtype was further characterized by altered expression of immune-related pathways. Molecular signatures measured in adenomatous PML may thus enhance our understanding of pathway dysregulation and mutational heterogeneity occurring during LUAD carcinogenesis, and implicate immunotherapeutic strategies to prevent their progression to cancer. Citation Format: Kelley Anderson, Linh Tran, Kostyantyn Krysan, William Wallace, Gregory Fishbein, Emily Green, Gang Liu, Hanqiao Liu, Erin Kane, Sarah Mazzilli, Steven Dubinett, Avrum Spira, Marc Lenburg, Jennifer Beane. Molecular subtyping of lung adenocarcinoma premalignant lesions identifies features associated with aggressive disease [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr PR007.

Abstract B022: A gradient expression of integrin αvβ6 expression in human cancer cells associates with TGFβ mediated immune evasion

Abstract Our previous results demonstrate an integrin-mediated mechanism of immune evasion in colorectal cancer. Increased levels of latent TGFβ in the experimental tumor environment led to the upregulation of the immune checkpoint marker PD-L1 in HCT116WT cells, facilitating escape from T-cell killing. This effect was rescued in immune co-cultures by pretreating cancer cells with the broad-spectrum integrin inhibitor GLPG-0187. To further elucidate this system, we investigated the TGFβ-integrin axis. Integrin αvβ6 is a major activator of the inert latent TGFβ into the active, immunosuppressive form. With its high specificity within malignant tissue, inhibition of αvβ6 poses an attractive therapeutic target for disrupting TGFβ signaling while avoiding the toxicities of systemic blockade. To identify an endogenous in vitro model of integrin αvβ6-expressing cancer, computational analysis of publicly available mRNA expression data was performed to determine cancer types with high αvβ6 upregulation. This investigation revealed a strong presence of αvβ6 in pancreatic, bladder, head and neck, and cervical cancer tissues. Further in-silico work identified the head and neck squamous cell carcinoma cell lines CAL-27 and FaDu, as well as the pancreatic adenocarcinoma line Capan-2, as suitable models for in vitro study. Flow cytometry of these cell lines was performed to determine the surface expression of integrin αvβ6, revealing that protein expression correlated with the mRNA data from the previous computational studies. Capan-2 and CAL-27 exhibited high levels αvβ6 of expression, while FaDu expressed moderate levels, and HCT116WT exhibited low αvβ6 expression. These cell lines were pretreated with latent TGFβ, simulating sequestered latent TGFβ in the extracellular matrix, and subsequently co-cultured with TALL-104 T-cells. Quantitative fluorescence microscopy revealed that resistance to killing by T-cells correlated with αvβ6 expression across cell lines. These findings are consistent with the hypothesis that upregulated expression of integrin αvβ6 functions as an immunosuppressive resistance mechanism by increasing tumor microenvironment levels of active TGFβ. Analysis of mRNA sequencing data from the TCGA Pancreatic Adenocarcinoma Firehose database revealed that integrin αvβ6 is a strong marker of poor prognosis in pancreatic adenocarcinoma. Future co-culture experiments will investigate the dynamics of TGFβ activation in the tumor microenvironment in relation to αvβ6 expression. Further in vivo studies can facilitate translation of our findings to therapeutic clinical trials. Citation Format: William J. MacDonald, Praveen R. Srinivasan, Vida Tajiknia, Maximilian Pinho-Schwermann, Connor Purcell, Brooke Verschleiser, Wafik S. El-Deiry. A gradient expression of integrin αvβ6 expression in human cancer cells associates with TGFβ mediated immune evasion [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B022.

Abstract B025: Spatially informed profiling of stage I lung adenocarcinoma reveals an extensive gene expression signature of vascular invasion

Abstract Microscopic vascular invasion (VI) is predictive of recurrence in stage I lung adenocarcinoma (LUAD) but is difficult to assess in resection specimens and cannot be accurately predicted prior to surgery. Thus, new biomarkers are needed to identify this aggressive subset of stage I LUAD tumors. To assess molecular and tumor microenvironment (TME) features associated with angioinvasive LUAD we profiled 171 resected stage I tumors with and without VI by RNA-seq, including 24 tumors by high-resolution spatial transcriptomics (stRNA-seq, 10x Genomics Visium). Visium capture areas were selected by an experienced thoracic pathologist to include invasive foci, tumor regions distal to foci, and tumors without invasive foci. We identified a molecular signature from the bulk RNA-seq discovery cohort (n=103) containing three gene expression clusters increased in VI+ stage I LUAD including genomic instability (C1), tissue remodeling (C2), and hypoxia (C3), and one increased in VI- (C4, immune surveillance). Analysis of the stRNA-seq revealed high inter-tumor patient heterogeneity, with most spots clustering by tumor identity, suggesting tumor-intrinsic properties. Scoring the stRNA-seq for VI signature clusters revealed that C2 was highly expressed outside the VI focus and declined as a function of distance toward it, while C1 and C3 increased. C2 was most strongly enriched in regions of desmoplastic but not normal stroma while C4 was enriched in normal-appearing adjacent lung and low-grade histologic patterns including lepidic and papillary. Given increased signature expression in regions of invasive tumors at a distance from VI foci, we leveraged the discovery cohort to develop a transcriptomic predictor of VI. We applied a nested cross-validation approach to select a machine learning model, which demonstrated an AUC of 0.86 for VI+ LUAD in our independent test set (n=60). In contrast, the predictor was unable to detect lymphatic invasion (LI) in VI- LUAD, suggesting a different molecular process may differentiate VI from LI. The VI-associated predictor increased across the spectrum of indolent to aggressive stage I LUAD histopathology and was predictive of recurrence-free survival, even in VI- LUAD, implying detection of pre-angioinvasive LUAD or VI foci that were missed during pathology review. Finally, we used RNA-seq data from multi-region sampling of stage I LUAD cases in TRACERx, a longitudinal study of NSCLC evolution, to evaluate the VI predictor’s robustness to intra-tumor heterogeneity (ITH). Across all tumors, we observed strong correlation between scores from randomly sampled tumor-matched regions (n=136; R = 0.91, p < 2.2e-16). The difference of scores between unmatched regions (inter-tumor heterogeneity) was significantly higher than matched regions (ITH) (p < 2.2e-16). Our study suggests that VI-associated gene expression extends from the site of intravasation and can be used to predict the presence of VI. This may enable the prediction of angioinvasive LUAD from biopsy specimens, allowing for more tailored treatment prior to surgery. Citation Format: Dylan Steiner, Lila Sultan, Jason Weis, Travis Sullivan, Emily Green, Hanqiao Liu, Sherry Zhang, Gang Liu, Avrum Spira, Sarah Mazzilli, Kimberly Christ, Eric Burks, Jennifer Beane, Marc Lenburg. Spatially informed profiling of stage I lung adenocarcinoma reveals an extensive gene expression signature of vascular invasion [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B025.

Abstract B019: The bone microenvironment's impact on the heterogeneity of resistance to proteasome inhibitors in multiple myeloma

Abstract Multiple myeloma (MM) is an osteolytic plasma cell malignancy that, despite being responsive to therapies such as proteasome inhibitors, frequently relapses. Understanding the mechanism and the niches where resistant disease evolves remains of major clinical importance. Cancer cell intrinsic mechanisms and bone ecosystem factors are known contributors to the evolution of resistant MM but the exact contribution of each is difficult to define with current in vitro and in vivo models. Using a novel bmathematical model that incorporates key cellular species of the bone ecosystem that control normal bone remodeling and, in MM, yields a protective environment we studied how, under therapy, the bone ecosystem contributes to the evolutionary dynamics of resistant MM under control and proteasome inhibitor treatment. Our results show that Environmentally-Mediated Drug Resistance (EMDR) might not be sufficient to explain resistant disease. But our results highlight the importance of EMDR in facilitating the emergence of resistance, and more importantly facilitating the survival of tolerant cells, allowing the tumor multiple routes for resistant disease. Our results also provide evidence that intervention with therapies targetting the bone ecosystem would significantly improve the ability of treatment to increase the time to progression for patients with multiple myeloma and, potentially, other diseases where EMDR plays a role. Citation Format: David Basanta, Ryan Bishop, Anna Miller, Conor C. Lynch, Matthew Froid, Ariosto S. Silva, Kenneth H. Shain. The bone microenvironment's impact on the heterogeneity of resistance to proteasome inhibitors in multiple myeloma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B019.

Abstract A023: Integrating mathematical modeling and in vitro experiments to measure ecological interactions in cancer

Abstract Integrating mathematical modeling and in vitro experiments to measure ecological interactions in cancer Maximilian Strobl, Dagim Tadele, Jeffrey Maltas, Rowan Barker-Clarke, Mina Dinh, Jacob Scott Over the past two decades it has become clear that tumours are complex and evolving ecosystems, in which different cancer cell populations interact with each other and with their non-tumour microenvironment. This understanding has given rise to the intriguing idea that we might be able to leverage these interactions to slow, or even revert, the expansion of resistant cells. In so-called “Adaptive Therapy” (AT) treatment is dynamically reduced to maintain the tumour burden tolerable whilst also preserving a pool of drug-sensitive cells which competitively suppress resistant cells. However, how strong and how common is this suppression, and how do we stratify patients who might benefit from AT from those who should receive standard-of-care continuous therapy? In this contribution, we present development of methodology and initial experimental results aimed at measuring the distribution of ecological interactions between drug-sensitive and resistant cells. Previously, we pioneered an assay with which to quantify ecological interactions in cancer in vitro. In the “Game Assay”, the two populations are co-cultured at a range of initial ratios in 2-D, and we record how the growth rate of each population is modulated by the relative abundance of the other. In the current study, we begin by using an agent-based mathematical model to characterize the accuracy and precision of this assay and how we may improve these by modifying different aspects of the experimental design (number of replicates, number of proportions) and analysis (regression technique, regression window). This shows that the choice of seeding ratios is important in being able to reliably identify interactions and we discuss guidelines for assay design. Upon identifying an interaction, a key next question is the underlying mechanism. Next, we thus use the model to investigate the degree to which we can use the assay to distinguish between local (e.g. contact inhibition) and non-local interactions (e.g. paracrine signaling). To conclude, we present preliminary experimental data in which we have used our optimized protocol to quantify the distribution of interactions between Osimertinib sensitive and resistant NSCLC cells (PC9). Our data show that while suppressive interactions do occur, they vary in their strength between different resistant lineages and can be attenuated in a frequency-dependent manner. Overall, we demonstrate how mathematical models can help in the design and interpretation of experimental assays and we contribute towards a more systematic and quantitative understanding of tumor ecology as a foundation for adaptive therapy. Citation Format: Maximilian A. R. Strobl, Dagim S. Tadele, Jeffrey Maltas, Rowan Barker-Clarke, Mina Dinh, Jacob Scott. Integrating mathematical modeling and in vitro experiments to measure ecological interactions in cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A023.

Abstract PR016: Topology of the tumor microenvironment: Integrating imaging, modeling, and topological data analysis

Abstract Cancer evolution involves multiple stochastic processes across many scales. In addition to genomic and transcriptomic changes, hallmarks of cancer within solid tumors include the malignant co-evolution of the tumor microenvironment. Although the evaluation of tissue architecture plays a key role in pathologist staging, the automated extraction of additional structural features remains far from translation to the clinic. However, one particularly effective novel methodology for extracting patterns from these clinical images is topological data analysis(TDA). TDA is a field that studies the shape of data and its connectedness using concepts from geometric and algebraic topology. Recent work using topological data analysis has allowed the extraction of image features. Such features have been associated with clinical outcomes in both the analysis of H&E images in prostate cancer and CT images in lung cancer. We set out to question how the topological features of tissues, including tumor and stromal composition, evolve during treatment and between original sites and metastasis. To analyze this we use a combination of tissue sections, imaging, and cell-automata models and persistent homology and cubical persistence techniques from TDA. We apply certain topological methods for the first time to tissue images (H&E) and to the output of spatial simulations. We focus specifically on analyzing the evolution of topological features in ovarian and lung cancer and compare our features to 2D and 3D simulations (HAL) with varying game theoretical interactions as well as comparing features across tissues from matched primary and metastatic sites. We find that several topological features are conserved between primary and metastatic sites and some feature evolution is replicated in simulation. Citation Format: Rowan Barker-Clarke, Jason M. Gray, Jacob G. Scott. Topology of the tumor microenvironment: Integrating imaging, modeling, and topological data analysis [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr PR016.

Abstract B007: Reconstruction of cellular phylogenies and analysis of gene expression variation in single cells

Abstract In cancer, somatic mutations occur continuously, causing cell populations to evolve. These somatic mutations result in the evolution of cellular gene expression patterns that can also change due to epigenetic modifications and environmental changes. By exploring the concordance of gene expression changes with molecular evolutionary trajectories of cells, we can examine the role of somatic variation on the evolution of gene expression patterns. In these endeavors, reliable inference of the evolutionary relationship of single cells is a key step. However, single-cell sequences contain many errors and missing bases, which necessitate advancing standard molecular phylogenetics approaches for applications in analyzing these datasets. We first show our new computational approach (TopHap+) that integratively applies standard phylogenetic optimality principles and patterns of co-occurrence of sequence variations to produce more expansive and accurate cellular phylogenies from single-cell sequence datasets. We then present Multi-Omics Concordance Analysis (MOCA) software to jointly analyze gene expressions and genetic variations from single-cell RNA sequencing profiles. MOCA outputs cells and genes showing convergent and divergent gene expression patterns in functional genomics. Citation Format: Sayaka Miura, Sudhir Kumar. Reconstruction of cellular phylogenies and analysis of gene expression variation in single cells [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B007.

Abstract A005: Quantification of cancer effect sizes of somatic variants reveals stage-dependent evolution in primary and metastatic skin cutaneous melanoma

Abstract The identities of driver and passenger genes in skin cutaneous melanoma (SKCM) have been intensively studied. However, the relative enhancement of cancer cell lineage survival and proliferation by SKCM driver mutations, as well as how these cancer effects vary with cancer stage, remains limited. The cancer effect size (also known as the scaled selection coefficient) quantifies the relative cancer cell lineage survival and proliferation in the context of each driver mutation. Using a diverse set of 999 primary and 1149 metastatic SKCM samples from multiple datasets, including The Cancer Genome Atlas (TCGA) and the AACR Project GENIE, we computed the cancer effect size of all recurrent mutations in primary (Stage I–III) and metastatic (Stage IV) tumors. Our analysis revealed that both primary and metastatic tumors were alike in their trinucleotide mutation profiles. However, the underlying gene mutation rates typically increased from primary to metastatic tumors as the latter group typically features ineffective DNA repair machinery. Additionally, the two cohorts noticeably differed in both magnitudes of effect sizes and in which genes specific somatic variants were most highly selected. Effect sizes of several driver mutations, most notably BRAF V600E, were orders of magnitude higher in metastatic tumors than in primary tumors. Furthermore, certain variants, particularly NONO R184L, were highly selected along the trajectory from normal tissue to primary tumors, yet new mutations were no longer selected between primary tumor resection and the development of resectable metastases. Conversely, for several variants, including H3F3B R18L, TP53 R273L, and PIK3CA R115P, selection was very strong only along the trajectory from resectable primary tumors to metastases. These findings illustrate the shifting evolutionary trajectory of tumors across cancer stages, underscoring the importance of early detection of SKCM and informing the development of targeted therapeutics aimed at specific stages of tumorigenesis. Citation Format: Rishi M Shah, Moein Rajaei, Jeff Mandell, Jeffrey P Townsend. Quantification of cancer effect sizes of somatic variants reveals stage-dependent evolution in primary and metastatic skin cutaneous melanoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A005.

Abstract PR011: Comprehensive longitudinal tracking of lung cancer evolutionary clonal dynamics during therapy using circulating tumour DNA

Abstract Background: Outgrowth of resistance cancer cell populations is a common mechanism of therapy failure in oncology. Effective personalised medicine relies on targeting of ubiquitous aberrations; however, tumours are highly heterogenous. Liquid biopsies have the potential to provide representative tumour sampling at regular intervals through disease course, but extensive interrogation of longitudinal clonal evolution through therapy has been lacking, particularly for patients on a curative disease pathway. Methods: We analysed 1102 plasma samples from 202 early-stage non-small cell lung cancer patients who underwent multiregional whole exome sequencing of primary tumour, relapse and, where available, comprehensive post-mortem samples. We designed personalised panels targeting a median of 200 mutations for each patient and sequenced cell-free DNA (cfDNA) around these variant sites to saturation. For four patients we designed personalised panels including all clonal and subclonal mutations discovered (up to 600). One of these four patients received a personalised neoantigen vaccine. We used ECLIPSE to determine the clonal composition of low tumour fraction ctDNA. Results: We found a directly proportional correlation between the cancer cell fraction (CCF) of subclones estimated in plasma and multiregional tissue samples at surgery. Lower plasma CCFs were observed for clonal illusion mutations from randomly selected single samples, one for each TRACERx patient, when compared to truly clonal mutations (P < 0.001, OR = 0.44), distinguishing appropriate therapeutic targets without multiregional sampling. In 24 patients with both cfDNA and tissue available at relapse we detected 28/29 relapse tissue subclones in the cfDNA with an additional 8 cfDNA unique subclones. We found a strong bias for cfDNA unique subclones to be present in only subset of cells (P = 0.008, OR = 5.5), consistent with insufficient tissue sampling. We observed longitudinal clonal dynamics which were often, but not always, associated with treatment. Metastatic competent subclones had higher CCFs in the primary tumour, as measured in pre-operative plasma (P < 0.001, OR = 4.5). Upon receiving a personalised neoantigen vaccine a patient had regression of clones harbouring a targeted EGFR mutation. However, copy number loss of the targeted EGFR mutation occurred in a liver lesion which grew over the treatment period on surveillance radiology alongside a novel ctDNA subclone. Upon cessation vaccine treatment the resistance-associated subclone and liver lesion underwent regression. EGFR mutant clones become dominant in the tumour thereafter suggesting a fitness cost associated with vaccine resistant EGFR mutation loss in this patient, measured using ctDNA. Conclusions: We find evidence that comprehensive tracking of subclonal mutations in plasma can accurately profile the clonal structure of tumours over time, revealing clonal dynamics in response to treatment and biological determinants of metastasis with the potential to guide personalised medicine. Citation Format: Alexander M. Frankell, Christopher Abbosh, Kristiana Grigoriadis, Emilia Lim, Selvaraju Veeriah, Sophia Ward, Tak Karasaki, James R. M. Black, Mariam Jamal-Hanjani, Nicholas McGranahan, Charles Swanton. Comprehensive longitudinal tracking of lung cancer evolutionary clonal dynamics during therapy using circulating tumour DNA [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr PR011.

Abstract IA011: The epigenetic-based liquid biopsy odyssey of bedside pancreatic cancer diagnostics

Abstract Pancreatic cancer (PC), often diagnosed at an advanced stage, urgently requires innovative early detection methods to improve patient outcomes. Presently, there is no approved screening method for pancreas cancer and its high-risk individuals (HRIs), as existing techniques rely on costly and invasive imaging modalities, which face challenges in detecting interval cancers. This study presents an innovative approach that utilizes liquid biopsy technology combined with blood-based DNA methylation biomarkers—ADAMTS1, BNC1, and LRFN5. These biomarkers demonstrate a remarkable 94% diagnostic accuracy for early-stage PC detection, providing a transformative strategy in precision medicine. In line with precision medicine principles, the potential application of this liquid biopsy-based DNA methylation panel for monitoring HRIs offers a promising path toward personalized surveillance. The creation of an affordable bedside test with exceptional diagnostic accuracy has the potential to revolutionize early detection strategies for PCs. Identifying early curable stage PC through this liquid biopsy technique aligns perfectly with precision medicine, promising increased patient survival. If successfully integrated into routine healthcare, early detection could significantly advance the fight against pancreatic cancer, paving the way for more effective and patient-centric diagnostic approaches. Citation Format: Nita Ahuja. The epigenetic-based liquid biopsy odyssey of bedside pancreatic cancer diagnostics [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr IA011.

Abstract B034: Order-of-mutation effects on cancer progression: Models for myeloproliferative neoplasm

Abstract In some patients with myeloproliferative neoplasms (MPN), two genetic mutations are often found, JAK2 V617F and one in the TET2 gene. Whether or not one mutation is present will influence how the other subsequent mutation affects the regulation of gene expression. When both mutations are present, the order of their occurrence has been shown to influence disease progression and prognosis. We propose a nonlinear ordinary differential equation (ODE) model and a Moran process models to explain the non-additive and non-commutative mutation effects on recent clinical observations of gene expression patterns, proportions of cells with different mutations, and ages at diagnosis of MPN. These observations consistently shape our modeling framework. Our key proposal is that bistability in gene expression provides a natural explanation for many observed order-of-mutation effects. We also propose potential experimental measurements that can be used to confirm or refute predictions of our models. Citation Format: Yue Wang, Tom Chou, Blerta Shtylla. Order-of-mutation effects on cancer progression: Models for myeloproliferative neoplasm [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr B034.