ABSTRACT Grade IV glioma, formerly known as glioblastoma multiforme (GBM) is the most aggressive and lethal type of brain tumor, and its treatment remains challenging in part due to extensive interpatient heterogeneity in disease driving mechanisms and lack of prognostic and predictive biomarkers. Using mechanistic inference of node-edge relationship (MINER), we have analyzed multiomics profiles from 516 patients and constructed an atlas of causal and mechanistic drivers of interpatient heterogeneity in GBM (gbmMINER). The atlas has delineated how 30 driver mutations act in a combinatorial scheme to causally influence a network of regulators (306 transcription factors and 73 miRNAs) of 179 transcriptional “programs”, influencing disease progression in patients across 23 disease states. Through extensive testing on independent patient cohorts, we share evidence that a machine learning model trained on activity profiles of programs within gbmMINER significantly augments risk stratification, identifying patients who are super-responders to standard of care and those that would benefit from 2 nd line treatments. In addition to providing mechanistic hypotheses regarding disease prognosis, the activity of programs containing targets of 2 nd line treatments accurately predicted efficacy of 28 drugs in killing glioma stem-like cells from 43 patients. Our findings demonstrate that interpatient heterogeneity manifests from differential activities of transcriptional programs, providing actionable strategies for mechanistically characterizing GBM from a systems perspective and developing better prognostic and predictive biomarkers for personalized medicine.
e18509 Background: The standard of care for acute myeloid leukemia (AML) is to pursue re-induction chemotherapy if ≥ 5% blasts are present on the nadir bone marrow (BM) biopsy after induction chemotherapy. However, some patients with indeterminate responses (IR) or with residual disease (RD) are unable to undergo re-induction, and some of these patients may still obtain a complete remission (CR). The objective of this analysis is to evaluate long-term outcomes at a large academic medical center for AML patients with IR or RD who did not receive early re-induction chemotherapy. Methods: Between 2002 and 2009, 233 adult patients with newly diagnosed AML received intensive induction chemotherapy at Wake Forest Baptist Medical Center. Decisions to forego or proceed with re-induction based on the 14-day (± 3 days) nadir BM biopsies were reviewed. Response to therapy was based on the nadir and recovery BM aspirates and biopsies. At the nadir, IR and RD were defined as 5 – 20% and > 20% blasts, respectively. CR was achieved if platelets ≥ 100,000/µL, absolute neutrophil count ≥ 1000/µL, and the recovery BM had < 5% blasts. Results: Of 233 patients, 90 (38.6%) had no evidence of leukemia on their nadir BM, 56 (24.0%) had < 5% blasts but features of the BM remained concerning for residual leukemia, 50 (21.5%) had an IR, and 37 (15.9%) had RD. Thirty patients with < 5% blasts, 36 patients with an IR, and 34 patients with RD underwent re-induction. Fourteen patients with an IR did not undergo re-induction, and 9 (64.3%) of these patients were still able to achieve a CR. Twenty-six patients with < 5% blasts on their nadir BM but with features concerning for residual leukemia did not undergo re-induction, and 23 (88.5%) of these patients achieved a CR. Only 3 patients with RD did not undergo re-induction, and none of these patients achieved a CR. Conclusions: Obtaining < 5% blasts on the nadir BM had 92.0% sensitivity in predicting a CR. However, ≥ 5% blasts on the nadir BM had only 38.1% specificity in predicting the failure of achieving a CR. Due to the toxicity associated with AML treatment, better criteria for identifying pathologically occult residual leukemia are urgently needed to help determine which patients need re-induction.
Abstract Glioblastoma (GBM) is a heterogeneous tumor made up of cell states that evolve over time. Here, we modeled tumor evolutionary trajectories during standard-of-care treatment using multimodal single-cell analysis of a primary tumor sample, corresponding mouse xenografts subjected to standard of care therapy, and recurrent tumor at autopsy. We mined the multimodal data with single cell SYstems Genetics Network AnaLysis (scSYGNAL) to identify a network of 52 regulators that mediate treatment-induced shifts in xenograft tumor-cell states that were also reflected in recurrence. By integrating scSYGNAL-derived regulatory network information with transcription factor accessibility deviations derived from single-cell ATAC-seq data, we developed consensus networks that regulate subpopulations of primary and recurrent tumor cells. Finally, by matching targeted therapies to active regulatory networks underlying tumor evolutionary trajectories, we provide a framework for applying single-cell-based precision medicine approaches in a concurrent, neo-adjuvant, or recurrent setting. Summary Inference of mechanistic drivers of therapy-induced evolution of glioblastoma at single cell resolution using RNA-seq and ATAC-seq from patient samples and model systems undergoing standard-of-care treatment informs strategy for identification of tumor evolutionary trajectories and possible cell state-directed therapeutics.
Abstract Glioblastoma is a heterogeneous tumor made up of cell states that evolve over time. We modeled tumor evolutionary trajectories during standard-of-care treatment using multimodal single-cell analysis of a primary tumor sample, corresponding mouse xenografts subjected to standard of care therapy, and recurrent tumor at autopsy. We mined the multimodal data with single cell SYstems Genetics Network AnaLysis (scSYGNAL) to identify a network of 52 regulators that mediate treatment-induced shifts in xenograft tumor-cell states that were also reflected in recurrence. By integrating scSYGNAL-derived regulatory network information with transcription factor accessibility deviations derived from single-cell ATAC-seq data, we developed consensus networks that regulate subpopulations of primary and recurrent tumor cells. Finally, by matching targeted therapies to active regulatory networks underlying tumor evolutionary trajectories, we provide a framework for applying single-cell-based precision medicine approaches in a concurrent, neo-adjuvant, or recurrent setting. Our proof-of-concept work herein provides the basis for the development of a modeling and analytical system that enables single-cell characterization of an individual patient’s tumor and inferred therapeutic vulnerabilities. Although further validation is required, in the form of in vivo studies of these putative druggable targets, our preliminary analysis and results suggest that systems biology techniques can be used to infer and predict therapeutic vulnerabilities that are either selected or induced during standard-of-care treatment. Ultimately, the information gathered from such systematic modeling and analysis of individual tumors may inform clinical treatment in a more targeted manner and enable a rational, tailored precision medicine that accounts for intratumoral cell heterogeneity.
Introduction: Anthracycline and trastuzumab chemotherapy related cardiac dysfunction (CTRCD) contributes to increased morbidity to breast cancer patients, yet predictors of CTRCD are not well defin...
Purpose: Objective documentation of airflow obstruction is often lacking inhospitalized patients treated for acute exacerbation of chronic obstructive pulmonary disease (AECOPD). The utility of spirometry performed in hospitalized patients to identify airflow obstruction, and thus a diagnosis of COPD, is unclear. Our aim was to compare inpatient spirometry, performed during an AECOPD, with outpatient spirometry. Methods: A retrospective analysis of data from patients enrolled in an AECOPD care plan was performed. As part of the plan, patients underwent inpatient spirometry to establish a COPD diagnosis and outpatient clinic spirometry within 4 weeks of hospital discharge to confirm it. Data analyzed included forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), slow vital capacity (SVC) and FEV1/ vital capacity (VC). Obstruction was defined by FEV1/VC<0.70. Results: A total of 159 patients (mean age 63.2 +/- 10.5 years) had corresponding in- and outpatient spirometry. The median days between inpatient and outpatient spirometry was 12 (interquartile range [IQR] 9-16). Inpatient spirometry had a sensitivity of 94%, specificity of 24%, positive predictive value of 83% and negative predictive value of 53% for predicting outpatient obstruction. The area under curve for using inpatient spirometry was 0.82. The mean difference between inpatient and outpatient FEV1 was 0.44 +/- 0.03 liters or 17.3 +/- 1.13 % predicted (p<0.0001) for FEV1. Conclusions: Inpatient spirometry accurately predicts outpatient airflow obstruction, thus providing an opportunity to identify patients admitted with suspected AECOPD who have no prior spirometric documentation.
