Eckel‐Passow, Jeanette E.

Abstract Establishing causal links between genetic polymorphisms and increased heritable risk of developing brain cancer is a major challenge. The non-coding single nucleotide polymorphism rs55705857 (A >G) is associated with a ~6-fold increased risk to develop IDH-mutant low-grade glioma (LGG). The rs55705857 G allele has a minor allele frequency of only ~5% in the general population but is found in ~40% of patients with IDH-mutant LGG and patients carrying risk-alleles are diagnosed on average 7-12 years earlier than those carrying non-risk A alleles. This makes rs55705857 one of the highest reported genetic associations with cancer, comparable with inherited BRCA1 gene mutations and the risk of developing breast cancer or other familial glioma genes such as NF1/2, CDKN2A or p53. To generate a LGG mouse model, we combined clonal activation of IDH1R132H with mutations of Trp53 and Atrx, which resulted in the development of LGG-like brain tumors in 25% of mice. Mutating the highly conserved, orthologous mouse rs55705857 locus to mimic the human risk allele dramatically accelerated tumor development from 463 to 172 days and increased penetrance to 75%. The resulting tumors exhibit elevated R-2-hydroxyglutarate levels, well-differentiated fibrillary neoplastic histology and metabolic rewiring, recapitulating histopathological and molecular hallmarks of human LGG. Mechanistically, we show that the rs55705857 locus resides within a brain-specific enhancer, which shows enhanced activity in IDH-mutant tumors. In addition, we found that the risk allele disrupts OCT2/4 binding, allowing increased interaction with the Myc promoter and increased Myc expression. The hyperactive chromatin status combined with the tissue specificity of this enhancer explains the cooperativity between mutant IDH and rs55705857 and why rs55705857 is associated specifically with IDH-mutant glioma, but not other cancers. Overall, we generated new LGG mouse models, which provide insights into the pathophysiology of this deadly disease and shed light into the heritable predisposition to LGG development.

AbstractObjectiveSatiety, defined as the duration of the sensation of fullness, is usually measured by validated visual analog scales (VAS) for appetite. Gastric function plays a key role in food intake regulation. However, the association between gastric emptying (GE) and VAS appetite is unknown.MethodsIn this cross‐sectional study, 134 participants (mean [SEM] age = 39 [0.8] years, mean [SEM] BMI = 38 [0.5] kg/m2, 67% females) completed simultaneous measurements of GE and VAS appetite. After a 320‐kcal meal, GE was measured by scintigraphy and appetite by validated 100‐mm VAS for 240 minutes. Satiation was defined as calories consumed to terminate meal and was measured by an ad libitum meal. GE, VAS, and ad libitum meal tests were measured on the same day. Percent of meal retention in the stomach, VAS area under curve (AUC0‐240 min), and overall appetite score (OAS) were calculated. Pearson correlation (ρ) determined the association of GE with VAS appetite and satiation. Appetite components were also analyzed by quartiles based on GE120 min.ResultsGE120 min was correlated with sensation of VAS hungerAUC(0‐240 min) (ρ = 0.24, p = 0.004), fullnessAUC(0‐240 min) (ρ = 0.16, p = 0.05), and OASAUC(0‐240 min) (ρ = 0.20, p = 0.02). Patients with rapid GE120 min had a mean increase in VAS hungerAUC(0‐240 min) by 32 mm/min (15.62%, p = 0.03) compared with normal/slow GE120 min.ConclusionsGE is associated with the sensations of appetite, and rapid GE is associated with increased appetite, which may contribute to weight gain.

Abstract Background Glioblastoma (GBM) is an incurable disease with few approved therapeutic interventions. Radiation therapy (RT) and temozolomide (TMZ) remain the standards of care. The efficacy and optimal deployment schedule of the orally bioavailable small-molecule tumor checkpoint controller lisavanbulin alone, and in combination with, standards of care were assessed using a panel of IDH-wildtype GBM patient-derived xenografts. Methods Mice bearing intracranial tumors received lisavanbulin +/−RT +/−TMZ and followed for survival. Lisavanbulin concentrations in plasma and brain were determined by liquid chromatography with tandem mass spectrometry, while flow cytometry was used for cell cycle analysis. Results Lisavanbulin monotherapy showed significant benefit (P < .01) in 9 of 14 PDXs tested (median survival extension 9%-84%) and brain-to-plasma ratios of 1.3 and 1.6 at 2- and 6-hours postdose, respectively, validating previous data suggesting significant exposure in the brain. Prolonged lisavanbulin dosing from RT start until moribund was required for maximal benefit (GBM6: median survival lisavanbulin/RT 90 vs. RT alone 69 days, P = .0001; GBM150: lisavanbulin/RT 143 days vs. RT alone 73 days, P = .06). Similar observations were seen with RT/TMZ combinations (GBM39: RT/TMZ/lisavanbulin 502 days vs. RT/TMZ 249 days, P = .0001; GBM26: RT/TMZ/lisavanbulin 172 days vs. RT/TMZ 121 days, P = .04). Immunohistochemical analyses showed a significant increase in phospho-histone H3 with lisavanbulin treatment (P = .01). Conclusions Lisavanbulin demonstrated excellent brain penetration, significant extension of survival alone or in RT or RT/TMZ combinations, and was associated with mitotic arrest. These data provide a strong clinical rationale for testing lisavanbulin in combination with RT or RT/TMZ in GBM patients.

Abstract Background Appropriately designed preclinical patient-derived xenograft (PDX) experiments are important to accurately inform human clinical trials. There is little experimental design guidance regarding choosing the number of PDX lines to study, and the number of mice within each PDX line. Methods Retrospective data from IDH-wildtype glioblastoma preclinical experiments evaluating a uniform regimen of fractionated radiation (RT), temozolomide (TMZ) chemotherapy, and concurrent RT/TMZ across 27 PDX lines were used to evaluate experimental designs and empirically estimate statistical power for ANOVA and Cox regression. Results Increasing the number of PDX lines resulted in more precise and reproducible estimates of effect size. To achieve 80% statistical power using ANOVA, experiments using a single PDX line required subsampling of 6 mice per PDX for each treatment group to detect a difference in survival of 135 days, and 9 mice per PDX to detect a difference of 100 days. Alternatively, a design that used 10 PDX lines had greater than 80% power to detect a difference of 135 days with a single mouse per PDX per treatment group, a difference of 100 days with 2 mice per PDX per treatment, and 35 days with more than 10 mice per PDX per treatment. Power for Cox regression was slightly smaller than ANOVA for very small experiments regardless of effect size and slightly higher than ANOVA for detecting a smaller effect size of 35 days difference in survival for moderate-to-large experiments. Conclusions Experimental designs using few mice across many PDX lines can provide robust results and account for inter-tumor variability.

Abstract Background Antibody drug conjugates (ADCs) targeting the epidermal growth factor receptor (EGFR), such as depatuxizumab mafodotin (Depatux-M), is a promising therapeutic strategy for glioblastoma (GBM) but recent clinical trials did not demonstrate a survival benefit. Understanding the mechanisms of failure for this promising strategy is critically important. Methods PDX models were employed to study efficacy of systemic vs intracranial delivery of Depatux-M. Immunofluorescence and MALDI-MSI were performed to detect drug levels in the brain. EGFR levels and compensatory pathways were studied using quantitative flow cytometry, Western blots, RNAseq, FISH, and phosphoproteomics. Results Systemic delivery of Depatux-M was highly effective in nine of 10 EGFR-amplified heterotopic PDXs with survival extending beyond one year in eight PDXs. Acquired resistance in two PDXs (GBM12 and GBM46) was driven by suppression of EGFR expression or emergence of a novel short-variant of EGFR lacking the epitope for the Depatux-M antibody. In contrast to the profound benefit observed in heterotopic tumors, only two of seven intrinsically sensitive PDXs were responsive to Depatux-M as intracranial tumors. Poor efficacy in orthotopic PDXs was associated with limited and heterogeneous distribution of Depatux-M into tumor tissues, and artificial disruption of the BBB or bypass of the BBB by direct intracranial injection of Depatux-M into orthotopic tumors markedly enhanced the efficacy of drug treatment. Conclusions Despite profound intrinsic sensitivity to Depatux-M, limited drug delivery into brain tumor may have been a key contributor to lack of efficacy in recently failed clinical trials.

Abstract Obesity is a multifactorial disease with a variable and underwhelming weight loss response to current treatment approaches. Precision medicine proposes a new paradigm to improve disease classification based on the premise of human heterogeneity, with the ultimate goal of maximizing treatment effectiveness, tolerability, and safety. Recent advances in high-throughput biochemical assays have contributed to the partial characterization of obesity's pathophysiology, as well as to the understanding of the role that intrinsic and environmental factors, and their interaction, play in its development and progression. These data have led to the development of biological markers that either are being or will be incorporated into strategies to develop personalized lines of treatment for obesity. There are currently many ongoing initiatives aimed at this; however, much needs to be resolved before precision obesity medicine becomes common practice. This review aims to provide a perspective on the currently available data of high-throughput technologies to treat obesity.

Abstract Introduction Tumefactive multiple sclerosis (MS) and glioblastoma (GBM) can be difficult to differentiate. Clinical presentations are nonspecific and may occur in patients with both diseases. Magnetic resonance imaging (MRI) scans are routinely used to generate an initial differential diagnosis. Although imaging features can help distinguish these entities, there are no pathognomonic signs, neither on traditional nor advanced MRI. Moreover, the ability to differentiate these entities depends on the radiologists’ experience. Ensuring the correct diagnosis has important therapeutic and prognostic implications, as the management of these diseases is very different. To date, a brain biopsy often represents the only option to obtain a diagnosis, but it may be inconclusive and could add unnecessary morbidity. We hypothesize that a machine-learning tool based on routinely performed MRI scans could guarantee a more reliable, fast, and reproducible diagnosis. Methods Only patients with a histological diagnosis were included in the study. Patients that underwent any form of radiation therapy before the surgery were excluded to avoid confounding factors due to treatment-related effects. A total of 99 MS and 257 GBMs were included, and preoperative T2-weighted images were collected. For each patient, the lesions' segmentation was performed, including both the contrast-enhancing and non-enhancing areas of the lesions. The segmentation masks were used to extract radiomic features from the T2-weighted sequences. All MRI images were standardized to the same mean and standard deviation before feature extraction and resampled to a voxel size of 1.0 × 1.0 × 1.0 mm. Features extraction was performed using Pyradiomics (v. 3.0.1) and included first order, shape, and texture features, totaling 100 features. The classifier consisted of a support vector machine (SVM) model. Patients were randomly assigned to the training and test sets with an 80-20% proportion for each class. To select the best hyperparameters for the SVM classifier, a grid-search analysis was performed on the training set using a nested cross-validation scheme (k=5). All experiments were performed using Python (v. 3.5) and the Scikit-learn library (v. 0.23.1). Results The mean accuracy and area under the ROC curve for the best set of hyperparameters on the validation set were 0.83 and 0.86, respectively. The weighted-average precision, recall, and F1 scores on the test set were 0.78, 0.76, and 0.77, respectively. Conclusions Radiomic features extracted from a routinely acquired MRI sequence showed promising results in differentiating tumefactive multiple sclerosis from glioblastoma, opening the possibility to obtain a reliable diagnosis without the need for a brain biopsy. Future experiments Future experiments will focus on using post-contrast T1-weighted images for the radiomic features extractions, testing multiple machine learning models and features selection algorithms, and including more diagnostic classes (i.e., metastasis, lymphomas). Citation Format: Gian Marco Conte, Jeanette E. Eckel-Passow, W. Oliver Tobin, Paul Decker, Daniel H. Lachance, Robert B. Jenkins, Bradley J. Erickson. Differentiation of tumefactive multiple sclerosis and glioblastoma using radiomics features extracted from magnetic resonance imaging and machine learning [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-023.

AbstractBackgroundLarge-scale genome-wide association studies (GWAS) have implicated thousands of germline genetic variants in modulating individuals’ risk to various diseases, including cancer. At least 25 risk loci have been identified for low-grade gliomas (LGGs), but their molecular functions remain largely unknown.MethodsWe hypothesized that GWAS loci contain causal single nucleotide polymorphisms (SNPs) that reside in accessible open chromatin regions and modulate the expression of target genes by perturbing the binding affinity of transcription factors (TFs). We performed an integrative analysis of genomic and epigenomic data from The Cancer Genome Atlas and other public repositories to identify candidate causal SNPs within linkage disequilibrium blocks of LGG GWAS loci. We assessed their potential regulatory role via in silico TF binding sequence perturbations, convolutional neural network trained on TF binding data, and simulated annealing–based interpretation methods.ResultsWe built an interactive website (http://education.knoweng.org/alg3/) summarizing the functional footprinting of 280 variants in 25 LGG GWAS regions, providing rich information for further computational and experimental scrutiny. We identified as case studies PHLDB1 and SLC25A26 as candidate target genes of rs12803321 and rs11706832, respectively, and predicted the GWAS variant rs648044 to be the causal SNP modulating ZBTB16, a known tumor suppressor in multiple cancers. We showed that rs648044 likely perturbed the binding affinity of the TF MAFF, as supported by RNA interference and in vitro MAFF binding experiments.ConclusionsThe identified candidate (causal SNP, target gene, TF) triplets and the accompanying resource will help accelerate our understanding of the molecular mechanisms underlying genetic risk factors for gliomas.

Abstract Background: The four most commonly-mutated genes in clear cell renal cell carcinoma (ccRCC) tumors are BAP1, PBRM1, SETD2 and VHL. And, there are currently 14 known RCC germline variants that have been reproducibly shown to be associated with RCC risk. However, the association of germline genetics with tumor genetics and clinical aggressiveness are unknown.Methods: We analyzed 420 ccRCC patients from The Cancer Genome Atlas (TCGA). Molecular subtype was determined based on acquired mutations in BAP1, PBRM1, SETD2 and VHL. Aggressive subtype was defined clinically using Mayo SSIGN score and molecularly using the ccA/ccB gene expression subtype. Publically-available Hi-C data were used to link germline risk variants with candidate target genes. Results: The 8q24 variant rs35252396 was significantly associated with VHL mutation status (OR=1.6, p=0.0037) and SSIGN score (OR=1.9, p=0.00094), after adjusting for multiple comparisons. We observed that, while some germline variants have interactions with nearby genes, some variants demonstrate long-range interactions with target genes.Conclusions: These data further demonstrate the link between rs35252396, HIF pathway and ccRCC clinical aggressiveness, providing a more comprehensive picture of how germline genetics and tumor genetics interact with respect to tumor development and progression.