Citations

Background: Cardiac resynchronization therapy (CRT) improves global systolic function in left ventricle. However, the impact of CRT on regional myocardial function is less clear. The aim of this study, therefore, is to evaluate regional changes in myocardial strain after CRT. Methods: Speckle tracking echocardiography (STE) was performed in 91 patients before and 6 month after CRT. Peak longitudinal strain was measured in 18 segments at base, mid and apex in three apical views. Left ventricular end-systolic volume (ESV) was measured by biplane Simpson’s method and reverse remodeling was defined as 10 % reduction in ESV at 6 month after CRT. Results: Peak strain improved in basal and mid segments of medial walls (basal inferoseptal, anteroseptal and inferior segments, p<0.05, mid inferoseptal and inferior segments: p<0.01) in entire group after CRT, with a greater degree of improvement in CRT responders than in nonresponders (Figure ). There was no significant improvement in free wall strain, with significant reduction in basal inferolateral segment after CRT (P=0.047) in entire group. Nonresponders had significant reduction in strain at the mid anterolateral segment (P=0.004) (Figure ). Strain of the apical segments also did not improve. Conclusions: CRT induces reverse remodeling by enhancing regional myocardial function in the septum, but suppresses the function in the free wall. Lack of reverse remodeling after CRT was characterized by the reduced function in lateral segments without improvement in septal function..

Background : Renovascular disease (RVD) and impaired renal function are important risk factors for cardiac disease, but it remained unclear whether selective recovery of renal function can reverse cardiac dysfunction. Aims: This study tested the hypothesis that improvement in renal function using selective intra-renal administration of endothelial progenitor cells (EPC) in swine renovascular hypertension would improve myocardial microvascular (MV) integrity. Methods: The functions of the myocardial MV (response to IV adenosine) and the stenotic kidney of 7 pigs were assessed in-vivo using multi-detector CT after 6 weeks of RVD and hypertension. Autologous EPC expanded in culture were then infused (10x10 6 cells) intra-renally, and in vivo studies repeated 4 wks later. Six normal pigs served as controls. Results : Glomerular filtration rate was lower in RVD compared to normal (47.9±10.1 vs. 70.8±4.3 mL/min, p<0.05) and improved after EPC (to 62.7±1.5 mL/min, p=NS vs. normal), as did renal blood flow. Mean arterial pressure remained elevated after EPC (120.7±11 and 129.4±6 vs. 99.9±4.7 mmHg in normal, p<0.05), and plasma renin activity was unchanged. Myocardial MV permeability-surface area increased in RVD in response to adenosine (0.009±0.001 to 0.011±0.002 AU, p=0.01), while MV volume fraction decreased (4.3±0.8 to 2.7±0.6%, p=0.01), indicating impaired MV barrier function and integrity. Contrarily, both remained unchanged in response to adenosine 4 wks after EPC (0.01±0.002 to 0.01±0.002 AU, p=0.24, and 3.5±0.8 to 3.4±0.5%, p=0.46, respectively), as did normal MV. Conclusion : A single intra-renal infusion of EPC that improved renal function in experimental RVD also preserved remote myocardial MV function, despite sustained systemic hypertension. These findings underscore the cross talk between renal and cardiac function, and the potential of selective renal EPC intervention to preserve both the kidney and heart in RVD. Renal and myocardial microvascular function change in normal and RVD before and after EPC treatment

Introduction: Whether echocardiographic (echo) markers of left ventricular (LV) remodeling and diastolic dysfunction contribute incremental and independent prognostic information to clinical risk markers in patients (Pts) with coronary artery disease and severe LV systolic dysfunction is unclear. We sought to determine which echo variables provide the greatest prognostic value in the Surgical Treatment for Ischemic Heart Failure (STICH) population. Methods: Pts enrolled in STICH for whom transmitral Doppler (E/A ratio) was available on a baseline echo interpreted by an echo core laboratory blinded to treatment and outcomes formed the analysis cohort. Comprehensive datasets to account for missing echo data were created by multiple imputation and the impact on all-cause mortality was determined with the Cox’s regression model. Results: E/A ratio could be measured in 1511 of the 2136 Pts enrolled in STICH. Amongst markers of diastolic dysfunction, E/A ratio was the most significant predictor of mortality (χ 2 41.05, p <0.001) with a non-linear, u-shaped, relationship. Mortality was lowest with E/A ratio = 1.0, and increased for E/A ratio <0.6 and >1.0 up to 2.3, beyond which there was no further increase in risk. The combination of larger LV end-systolic volume index (LVESVI), low or high E/A ratio, and mitral regurgitation severity grade, had highly significant incremental negative effects on mortality (χ 2 69.65, p<0.001) when added to a multivariable model with clinical risk markers. Overall, creatinine (χ 2 30.00, p <0.001), followed by LVESVI (χ 2 27.26, p<0.001), age, and E/A ratio (χ 2 12.46, p<0.001) were among the most significant predictors of mortality and accounted for 74% of the total prognostic information. LVESVI and E/A ratio were stronger predictors of poor prognosis than New York Heart Association (NYHA) functional class, hemoglobin, diabetes, stroke, or atrial fibrillation. Conclusions: Echo markers of advanced LV remodeling and diastolic dysfunction add incremental prognostic value to clinical risk markers and are more predictive of poor prognosis than advanced NYHA functional class or anemia. LVESVI and E/A ratio outperformed other echo markers and should be considered standard in assessing risk in Pts with ischemic LV dysfunction.

Background: Atrial fibrillation (AF) is the most common cardiac arrhythmia in clinical practice, causes substantial morbidity, and is associated with an excess risk of mortality. However, information on contemporary time trends in survival after AF is lacking. Thus, we aimed to determine whether survival after incident AF has changed over the last decade in a community-based cohort of 3413 individuals with incident AF between 2000 and 2010. Methods: Olmsted County, MN residents 18 years of age and older with a first-ever AF or atrial flutter event between 2000 and 2010 were identified using inpatient and outpatient diagnostic codes and electrocardiograms. Deaths from any cause were ascertained through August 31, 2012 and time trends in survival after AF were examined using Kaplan-Meier curves and Cox proportional hazards regression. Results: Among 3413 individuals with incident AF (52% men, age range 18-104) between 2000 and 2010, 1561 deaths were observed over a median follow-up of 3.4 years. Survival did not differ by year of AF diagnosis (figure). In addition, the proportion of cardiovascular disease-related deaths compared to non-cardiovascular deaths did not differ over time. After adjustment for age, sex, and Charlson comorbidity index, the hazard ratios for all-cause mortality were 1.09 (95% CI 0.97-1.23) for those diagnosed with AF in 2004-2007 vs. 2000-2003 and 1.04 (95% CI 0.90-1.12) for 2008-2010 vs. 2000-2003. Furthermore, the risk of death did not differ by sex after adjustment for age and comorbidity (HR 1.00, 95% CI 0.90-1.11 for men vs. women). Conclusions: In the community, survival after AF has remained constant over the last decade despite substantial changes in the recommendations for treatment of AF. This underscores the importance of identification of prognostic factors and continued surveillance of outcomes in AF, as well as a better understanding of how to optimize the management of AF to improve outcomes in these patients.

Introduction: Idiopathic Dilated (DCM) and hypertrophic cardiomyopathies (HCM) are leading causes of sudden cardiac death. The incidence and prevalence of both DCM and HCM were originally estimated in Olmsted County, MN over 1975-1984. Incidence was 6.0 and prevalence was 36.5 per 100,000 person-years, for DCM; with incidence and prevalence rates of 2.5 and 19.7 per 100,000 person-years for HCM. This study aims to determine the incidence and mortality rates of individuals diagnosed with HCM and DCM in Olmsted County, MN from 1984 to 2016. Methods: The Rochester Epidemiology Project is a validated long term medical record linkage system that collects and links medical information for the approximately 145000 residents of Olmsted County, Minnesota. Using this database and the Electronic Health Record systems at Mayo Clinic and Olmsted Medical Center in Rochester, MN, we manually reviewed and validated the medical records of 4440 patients. This cohort includes all patients with any mention of a cardiomyopathy based on ICD and HICDA codes during the period 1984 to 2016. Data were summarized with mean and standard deviation, frequency and percentage and compared between groups using two sample t-test for continuous data and chi-square for proportions. Survival was computed using Kaplan-Meier methods and compared to expected survival based on age and gender. Incidence was computed using Olmsted County population numbers during the time period and adjusted to the US 1980 white population to allow for comparison with the previous study. Results: The calculated crude incidence rate for DCM and HCM are 10.1 and 7.8 per 100,000 respectively. Incidence rates adjusted for comparison to US 1980 white population (age and sex adjusted) are 10.4 (95% CI: 9.3-11.4) and 8.0 (95% CI: 7.1-8.8) per 100,000 person-years, respectively. The observed number of deaths in this cohort (HCM and DCM) was 420 vs 194 expected. Conclusions: Calculated incidence, both crude and adjusted rates are higher than previously estimated rates. The crude incidence rate for HCM is almost 4-fold higher than that for DCM is 2.4 times higher than the original estimate by Codd et al. Individuals with cardiomyopathies were found to have a higher overall mortality rate than the general population with an observed to expected hazard ratio of 1.73 (95% CI 1.49-2.01, p<0.001) for DCM and 1.44 (95% CI 1.21-1.71, p<0.001) for HCM. Our findings suggest that previous studies may have underestimated the true burden of HCM and DCM in the population which has important implications for healthcare provision and cost.

Background: Current guidelines recommend oral anticoagulation for the prevention of stroke in most patients with atrial fibrillation (AF). Direct oral anticoagulants (DOACs) are associated with lower risk of ischemic stroke and intracranial bleeding than warfarin and, therefore, may reduce risk of dementia among patients with AF. Hypothesis: AF patients initiating DOACs for stroke prevention will have lower risk of dementia than patients initiating warfarin. Methods: We used data from two US healthcare claim databases, MarketScan (2007-2015) and Optum Clinformatics (2009-2015). Using validated algorithms, we identified patients with AF who initiated oral anticoagulation. Dementia was defined as having an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code of 290.xx, 294.xx, or 331.0. Comorbidities and use of other medications were defined based on inpatient, outpatient, and pharmacy claims. We performed a series of head-to-head comparisons of different oral anticoagulants (warfarin, dabigatran, rivaroxaban, and apixaban) in propensity score-matched cohorts. In each database, multivariable-adjusted Cox regression models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for incident dementia for each comparison of propensity score-matched cohorts, adjusting for age, sex, comorbidities, use of other medications, and markers of socioeconomic status (in the Optum database only). We meta-analyzed database-specific results using a random-effects model. Results: The analysis included 307,099 AF patients from the MarketScan database and 161,346 from the Optum database, of which 6,572 and 4,391 respectively had a diagnosis of incident dementia. Mean follow up ranged between 0.7 to 2.3 years and incident of diagnosed dementia between 7 to 14 cases per 1000 person-years across different oral anticoagulant initiator groups. Patients initiating DOACs had a lower risk of incident dementia than those initiating warfarin (dabigatran: HR 0.85, 95%CI 0.71, 1.01; rivaroxaban: HR 0.85, 95%CI 0.76, 0.94; apixaban: HR 0.80, 95%CI 0.65, 0.97). There were no differences in risk of incident dementia comparing DOAC user groups (dabigatran vs. rivaroxaban: HR 1.02, 95%CI 0.79, 1.32; dabigatran vs. apixaban: HR 0.92, 95%CI 0.63, 1.36; apixaban vs. rivaroxaban: HR 1.01, 95%CI 0.86, 1.19). Conclusions: Patients with AF initiating DOACs experienced lower rates of incident dementia than warfarin users. No obvious benefit was observed for any particular DOAC in relation to dementia rates.

Background: Oral anticoagulants (OACs) are recommended for AF patients for the prevention of thromboembolic events, including stroke. Stroke risk stratification scores (e.g., CHA 2 DS 2 -VASc) are used to tailor therapeutic recommendations for patients in different risk groups. However, these were derived before the advent of direct OACs. At present, there is no tool that estimates a patient’s stroke risk, given their individual characteristics, by type of OAC. Methods: We used healthcare utilization data from two independent US databases (MarketScan and Optum) to construct and validate a predictive model of ischemic stroke in patients with AF initiating OACs. Patients with non-valvular AF initiating OACs were identified from the MarketScan data for years 2007-2015. Using bootstrapping methods and backward selection of 44 candidate variables, we developed a model which selected variables predicting stroke. The final model was validated in patients with non-valvular AF in the Optum Clinformatics database in the period 2009-2015. Results: Among 135,523 patients with AF initiating OAC in the MarketScan dataset, 2,028 experienced an ischemic stroke after anticoagulant initiation. The stepwise model identified 15 variables (including type of OAC) associated with ischemic stroke (Table). The discrimination (c-statistic) of the model was adequate [0.68, 95% confidence interval (CI) 0.66-0.70], showing excellent calibration (χ 2 = 7.7 p=0.57). The model was then applied to the 84,549 AF patients in the Optum data set (1408 stroke events). The model showed similar discrimination (c-statistic 0.67, 95%CI 0.65-0.70) and calibration (χ 2 = 12.8 p=0.17). However, previously-developed predicted models had similar discriminative ability (CHA 2 DS 2 -VASc 0.67, 95%CI 0.65-0.68; ATRIA 0.67, 95%CI 0.65-0.68) Conclusion: A novel model using extensive administrative healthcare data for the identification of patients at higher risk of ischemic stroke by type of anticoagulant did not perform better than established simple models.

Abstract Background: Conventional prognostic factors for LBCL were not associated with outcomes in the pivotal ZUMA-1 study of axi-cel in relapsed LBCL (Neelapu et al. NEJM. 2017); however, other attributes like chimeric antigen receptor (CAR) T-cell fitness and composition (CCR7+CD45RA+ T cells), reduced preTx tumor burden, and immune tumor microenvironment (TME) with presence of activated CD8+PD-1+LAG-3+/-TIM-3- T cells were associated with efficacy (Locke et al. Blood Adv. 2020; Galon et al. ASCO 2020. #3022). Here, we evaluated preTx immune cell phenotypes in premanufacturing apheresis (premfg aph) material, comprising peripheral blood mononuclear cells, to determine associations with product attributes, immune TME features, and clinical efficacy in ZUMA-1. Methods: Evaluable samples from patients (pts) in Phase (Ph) 1 and Ph2 Cohorts (C) 1-3 were analyzed (NCT02348216; Ph1 and Ph2 C1+2, ≥2-y follow-up; C3, ≥6-mo follow-up). Memory T, myeloid, NK, NKT, and B cells in premfg aph material (n=101, excluding C3) were characterized by flow cytometry (FC). PreTx immune TME was analyzed by multiplex IHC (n=18) and gene expression analysis (n=30) as previously described (Rossi et al. AACR 2018. #LB-016; Galon et al. ASCO 2020. #3022). CAR T-cell fitness was analyzed by doubling time, viability during manufacturing, and product T-cell phenotypes by FC (n=145). Associations between these covariates, and with routine hematology tests, were performed by Spearman rank correlation or Wilcoxon tests. Effects on survival were assessed by Kaplan-Meier with optimized cutpoint selection. Results: The percentage (%) of naive (CCR7+CD45RA+) T helper (Th; CD4+CD127+CD25low) cells coexpressing CD27 and CD28 (median, 1%; range, 0.01%-15.8%; IQR, 0.3%-3.8%) in aph associated positively with axi-cel efficacy. The % of intermediate monocytes (IMs; CD14+CD16+; median, 1.8%; range, 0.003%-16.7%; IQR, 1%-3%) in aph associated negatively with efficacy. The % of circulating CD27+CD28+ naive Th cells associated positively with an enriched preTx immune TME T-cell signature, % CCR7+CD45RA+ product T cells, objective response rate, PFS, and OS. An increased % of IMs associated directly with negative predictive markers (preTx serum levels of LDH, IL-6, and CRP) and inversely with TME T-cell signature, PFS, and OS. The premfg ratio of CD27+CD28+ naive Th cells/IMs associated directly with CAR T-cell expansion and efficacy. Conclusions: This work points to a link between the pre-existing state of the immune system, reflected in premfg aph, and immune TME, as well as product attributes influencing axi-cel efficacy in LBCL. These data bear practical implications towards the development of predictive biomarkers for axi-cel efficacy. [JB and JC contributed equally.] Citation Format: Justin Budka, Justin Chou, Vicki Plaks, Francesca Milletti, Zixing Wang, Frederick L. Locke, Sattva S. Neelapu, David B. Miklos, Caron A. Jacobson, Lazaros J. Lekakis, Yi Lin, Armin Ghobadi, Zahid Bashir, Nathalie Scholler, Jérôme Galon, John M. Rossi, Adrian Bot. Pretreatment (PreTx) immune cell phenotypes in peripheral blood associated with the tumor immune contexture, product attributes, and durable clinical efficacy in patients with large B-cell lymphoma (LBCL) treated with axicabtagene ciloleucel (axi-cel) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr CT166.

Abstract Introduction: The irreversible pan-HER inhibitor NER showed modest single agent activity for HER2mut MBC in Part I of MutHER trial. In Part II, we hypothesized that (1) N+F would improve activity in estrogen receptor positive (ER+) HER2mut MBC due to ER-HER2 crosstalk and (2) dual HER2 blockade by adding trastuzumab at disease progression (PD) could overcome resistance. Methods: Pts with ER+HER2mut MBC were enrolled to 2 cohorts (FUL treated or naive) to receive N+F with diarrhea prophylaxis. ER- pts received NER in an exploratory ER- cohort. Trastuzumab was added at PD if approved by insurance. Simon's Minimax 2-stage phase II design with the primary endpoint of clinical benefit rate (CBR: rates of complete/partial response [CR/PR] plus stable disease [SD] &gt;24 weeks [wks]), with anticipated vs null hypothesis being CBR of 55% vs 35% (FUL treated) or 65% vs 40% (FUL naïve) with 80% power, 1 sided 0.05 alpha, was used. Secondary endpoints included progression free survival (PFS) and adverse events (AEs). Serial blood samples were analyzed for circulating tumor DNA (ctDNA) by Guardant360 for concomitant mutations, HER2mut variant allele frequency (VAF) dynamics, and resistance mechanisms. Results: Between Sep. 2015 and Oct. 2020, 40 pts with HER2mut MBC were enrolled, completing the 1st stage of each ER+ cohort. 35 pts (21 FUL treated, 10 FUL naïve, 4 ER-) were evaluable for response, with median age 63 (35-82) years, 3 (0-12) prior MBC regimen, lobular BC in 13 (37%) and visceral mets in 32 (91%) pts. 21 (68%) ER+ pts had prior CDK4/6 inhibitor. All but 1 pt has come off study due to PD. Table 1 shows the efficacy by cohort. Further enrollment is closed per protocol. Adding trastuzumab at PD induced CB in 4 (3 PR, 1 SD≥24 wks) of 5 pts (1 ER-, 4 ER+), with PFS 28 (95% CI 18~NA) wks. Common AEs across cohorts were diarrhea (G3 21%) and fatigue (G3 5%). No G4 AEs. ctDNA HER2mut was detected in 72% (23/32) baseline (BL) samples tested. In pts with paired samples, HER2mut VAF decreased at C1D15/C2D1 from BL in 75% (15/20) and rose in 89% (16/18) at PD. Acquired HER2mut, including the T798I gatekeeper mutation, were detected in 2 pts at PD. Mutations in TP53 (53%), PIK3CA (43%), and CDH1 (35%) were common, but none significantly associated with PFS in all or ER+ pts. Conclusions: NER, or N+F, is active for HER2mut MBC with good tolerability. Adding trastuzumab at PD induced further response, supporting dual HER2 blockade for HER2mut MBC. Table 1.EfficacyCohortFUL treatedFUL naïveER-Best Response, n evaluablen = 21n = 10n = 4CR, n100PR, n431SD (≥ 24 wks), n300SD (&lt; 24 wks), n1030PD, n343CBR, n with CB/total n evaluable, % (95% CI)8 of 20*, 40% (19~64%)3 of 10, 30% (7~65%)1 of 4, 25% (0.6~81%)mPFS (95% CI), wks, ITT (n)24 (16~31) wks, (n = 24)20 (8~NA) wks, (n = 11)8.5 (8~NA) wks, (n = 5)*20 of 21 pts are evaluable for CBR in the FUL treated Cohort as 1 pt had SD as best response and treatment is still ongoing. ITT (intent to treat) population is used for mPFS estimate. Citation Format: Cynthia X. Ma, Jingqin Luo, Rachel A. Freedman, Timothy Pluard, Julie Nangia, Janice Lu, Frances Valdez-Albini, Melody Cobleigh, Jason Jones, Nancy U. Lin, Eric Winer, P. Kelly Marcom, Shana Thomas, Jill Anderson, Brittney Haas, Kimberly M. Hamann, Richard Bryce, Alshad S. Lalani, Lisa Carey, Matthew Goetz, Feng Gao, Gretchen Kimmick, Mark Pegram, Matthew J. Ellis, Ron Bose. A phase II trial of neratinib (NER) or NER plus fulvestrant (FUL) (N+F) in HER2 mutant, non-amplified (HER2mut) metastatic breast cancer (MBC): Part II of MutHER [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr CT026.

Abstract Despite the well-established role of recurrent gene fusions as oncogenic drivers, current practices for characterizing and interpreting gene fusion events in clinical testing and in biomedical literature are inconsistent. From the conceptual definition of gene fusions to the salient elements that characterize these alterations, a lack of community-driven standards for the curation of gene fusions has resulted in a disparate landscape of fusion representations and supporting tools. Consequently, the evidence-based clinical evaluation of gene fusions requires extensive expert review for accurate interpretation of observed gene fusions with respect to putative evidence from biomedical literature. Furthermore, the lack of these standards inhibits the interoperability of tools, resources, and pipelines - impeding data sharing and downstream utility.To address these challenges, a cross-consortia initiative between the Variant Interpretation for Cancer Consortium and ClinGen was formed to develop a standard operating procedure (SOP) for the curation of gene fusions. The SOP is under development by an international and diverse set of experts in the representation, detection, and clinical interpretation of gene fusions. Participating stakeholders across academic, government, and industry sectors showcased challenges and solutions, and participated in community surveys and discussions to define and develop the SOP for this diverse class of alterations.An initial result of this effort was the precise molecular definition of genomic events and features constituting gene fusions. We distinguish these from similar but distinct classes of structural alterations through clinically-relevant examples. Next, we discuss our findings on community practices around the description and evaluation of gene fusions. We provide our recommendations for characterization and representation of gene fusions from these practices, and compare these recommendations to existing variant representation standards and formats (e.g. HGVS variant nomenclature). We also discuss the concurrent application of formats for standardized human- and machine-readable representations of gene fusion events.We conclude with discussion of the salient elements to enable rapid, scalable, and consistent evaluation of fusions curated from the biomedical literature. Recommendations are provided for the standardized capture of these elements to enable both intuitive and precise characterization of this diverse class of alterations in clinical reporting and literature. In summary, we provide a clinical-practice driven framework and nomenclature for gene fusions, including recommendations for human readability, computational precision, and data integrity within the SOP. This work is a substantial advancement towards standardized communication, investigation, and sharing of gene fusion data across clinical and research domains and specialties. Citation Format: Alex H. Wagner, Ioannis S. Vlachos, Dmitriy Sonkin, Panieh Terraf, Chimene Kesserwan, Andrea Sboner, Thomas Coard, Christian Reich, Deborah I. Ritter, Peter Horak, Ying S. Zou, Anna Tanska, Aaron M. Berlin, Anna Lu, Daniel Cameron, Heather E. Williams, Wan-Hsin Lin, Gokce Toruner, Arpad Danos, Jason Saliba, Huiling Xu, Xinjie Xu, Georgina Ryland, Michele Ceccarelli, Liying Zhang, Sarah Rapisardo, Catherine Rehder, Xuelu Liu, Aparna Pallavajjala, Nicole Park, Laveniya Satgunaseelan, Kristy Lee, Jie Liu, Obi Griffith, Robert R. Freimuth, Albrecht Stenzinger, Linda B. Baughn, Michael Baudis, Jennifer Lee, Marilyn Li, Angshumoy Roy, Gordana Raca. A standard operating procedure for the curation of gene fusions [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 449.