Citations

Abstract Clinical management of IDH wildtype glioblastoma (GBM) and tumefactive multiple sclerosis (tMS) is drastically different. GBM requires maximal safe resection followed by chemoradiation, while tMS outcome is worsened by surgery and radiotherapy. Noninvasive methods are needed to help with accurate diagnosis of tumor and non-tumor etiologies. To develop an MRI-based classification model, tMS subjects diagnosed prior to January 1, 2020, were matched to tMS by age at diagnosis, sex, index MRI date, and 2D/3D acquisition. Inclusion criteria included one cm minimal lesion size and pre-operative post-contrast T1 and T2 images available for analysis. A 3D-DenseNet121 was used to develop a classification model using prespecified parameters: 650 epochs, batch size 16, learning rate 10-3, cross-entropy loss, and AdamW optimizer. The stopping rule was defined as three sequential differences in epoch cross-entropy loss <0.02. Models were developed using both T1gd and T2, as well as from only T1gd and only T2. Training included 220 subjects (110 GBM, 110 tMS). A 2-stage validation design was used, which included both retrospective and prospective cohorts. Stage 1 consisted of 272 retrospective GBM (diagnosed prior to January 1, 2020). Stage 2 consisted of 69 and 34 prospective (diagnosed after January 1, 2020) GBM and tMS, respectively. External validation on the 272 retrospective GBM demonstrated accuracy of 91%, 84%, and 78% for T1gd+T2, T1gd only, and T2 only, respectively. External validation on the 69 prospective GBM demonstrated an accuracy of 87%, 64%, and 67% for T1gd+T2, T1gd only, and T2 only, respectively. The 34 prospective tMS demonstrated accuracy of 76%, 76%, and 82% for T1gd+T2, T1gd only, and T2 only, respectively. This shows the feasibility of deep learning to aid in differential diagnosis of brain lesions. Future work entails the integration of germline variants into the classification model, including variants associated with the risk of glioma or MS.

Abstract Acquired tumor alterations are part of the WHO diagnosis of glioma. However, germline predisposition to these alterations is unknown. GWAS in European and Asian populations identified germline variants in 30 regions, many in or near genes that also acquire alterations, suggesting that interactions between inherited and acquired genetics are involved in glioma development. To evaluate such interactions, we used 3686 glioma cases and 1889 controls. A case-control design and multinomial logistic regression were used to identify germline variants associated with development of TERT-promoter mutant gliomas, generating two interesting observations. First, TERT germline variants were associated with predisposition to IDHwt glioma that have a TERT promoter mutation, but not with IDH-mutant codeleted gliomas that have a TERT promoter mutation. Second, two novel regions were observed in this more homogeneous subgroup of TERT mutated tumors: variants in TXNDC11 (OR=2.66, p=5.7x10-8) and RUVBL2 (OR=3.1, p=5.8x10-8). High tumor expression of TXNDC11 is associated with poor prognosis in glioma and RUVBL2 interacts with TERT to affect telomerase activity. Similar analyses are being performed on additional clinically relevant acquired alterations. In parallel, case-case GWAS analyses were performed to evaluate germline variants associated with other tumor alterations. Comparing IDH-mutant versus IDHwt stratified by rs55705857 genotype, the most significant variants were PHLDB1 (p=1.2x10-13) and D2HGDH (p=2.1x10-10), highlighting the importance of these two variants independent of CCDC26 for the development of IDH-mutant glioma. As a parallel approach, we utilized recursive partitioning and regression trees (rpart) to classify IDH mutation status using the 30 known variants. The top four branches of the regression tree included CCDC26, D2HGDH, PHLDB1, and MAML2, as well as two Asian variants (RAB27A, CYP4F12). This highlights the importance of utilizing variants discovered in non-European populations in analyses. Overall, understanding these interactions are necessary to develop therapeutics, stratify clinical trials, and to develop relevant tumor models.

Abstract Histologic features, presence of CDKN2A/B homozygous deletion and methylation subtype are strong determinants of outcome and of therapy in IDH-mutant gliomas. We were interested in learning if there are additional markers of outcome in these tumors. To address this question, we assembled a cohort of 963 IDH-mutant gliomas with overall survival data (median follow-up alive 2.7 years, median follow-up deceased 3.9 years). Array-based copy number data, targeted NGS, germline genotyping data and Illumina EPIC array methylation data were available on 872, 226, 542, and 401 patients, respectively. Morphologic type (oligodendroglioma versus astrocytoma, grade (2/3 vs 4) and methylation subtype (A_IDH_LG vs A_IDH_HG) were all significantly associated with survival (p <0.001). There was no significant difference in survival between grade 2 vs 3 IDH-mutant astrocytomas (p=0.53 adjusting for age and sex). However, subjects with the risk allele (G) for rs55705857 had a better overall survival than subjects with the non-risk allele (A) in IDH-mutant oligodendroglioma (AG/GG vs AA, median overall survival 21.6 vs 15.7 years, p=0.048 after adjusting for grade, age, and sex). Conversely, rs55705857 was associated with a markedly worse survival in grade 2 IDH-mutant astrocytomas (AG/GG vs AA, median overall survival 8.6 vs 19.6 years, p=0.007 after adjusting for age and sex). Despite similar sample sizes and number of deaths, there was no significant difference in overall survival by rs55705857 genotype in grade 3 or 4 IDH-mutant astrocytomas. Upon further functional analyses, TLX1 and GALNT17 were found to be differentially methylated in IDH1-mutant astrocytomas and oligodendrogliomas, respectively, when stratified by rs55705857 risk allele. This is the first observation of a germline characteristic (e.g. rs55705857) predicting survival in low-grade IDH-mutant astrocytoma. While this observation requires validation, it suggests that germline genotype may be practice changing and inform results of trials such as CATNON and CODEL.

2071 Background: Brigimadlin (BI 907828) is a potent small molecule inhibitor of the p53-MDM2 pathway currently in a phase 0/1 clinical trial in combination with radiotherapy in newly diagnosed MGMTunmethylated glioblastoma (GBM). To inform development of brigimadlin for GBM, we developed a pharmacokinetic (PK)/ pharmacodynamic (PD)/ efficacy model using GBM patient derived xenografts (PDXs). Methods: Brigimadlin PK parameters were evaluated in FVB wild-type and knockout (Mdr1a/b-/-Bcrp1-/-) mice and in athymic nude mice. Drug binding was measured by rapid equilibrium dialysis, and drug levels quantified by LC/MS-MS. Dose-response relationships were evaluated in subcutaneous and orthotopic PDXs across a range of doses. Response to brigimadlin (2 mg/kg weekly) +/- RT (2 Gy x 10 fractions) was evaluated in orthotopic PDXs. Results: Brigimadlin significantly delayed median time to regrowth of GBM108 ( MDM2amplified) subcutaneous tumors (vs. placebo; p<0.05) at dose levels of 0.25 mg/kg (1.3-fold), 0.5 mg/kg (1.5-fold), 1 mg/kg (2.6-fold), and 2 mg/kg (5.3-fold). A single dose of brigimadlin resulted in corresponding intra-tumoral drug levels 24 hours post-dose: 84 nM (0.25 mg/kg), 175 nM (0.5 mg/kg), 398 nM (1 mg/kg) and 924 nM (2 mg/kg) and dose-dependent upregulation of p53 target genes (p21 and PUMA). In contrast, GBM14 ( MDM2non-amplified) showed regrowth delay (1.6-fold) only at 2 mg/kg (p=0.008). Brigimadlin was highly bound (fraction unbound = 0.0006 in brain; 0.0017 in GBM tumor tissue), and CNS distribution was limited by efflux; in wild-type mice Kp,uu_brain = 0.002 ± 0.003; in Mdr1a/b-/-Bcrp1-/- mice Kp,uu = 0.043 ± 0.025. Efficacy of brigimadlin was compared in orthotopic GBM108 PDXs grown in athymic nude mice vs. immuno/efflux-deficient (Rag2-/-Mdr1a-/-Bcrp1-/-) mice. In nude mice, median survival was extended at 2 mg/kg brigimadlin (1.4-fold, p=0.009), with corresponding intra-tumoral drug levels of 29 nM. In efflux deficient mice, median survival was extended (p<0.05) at 0.25 mg/kg (1.5-fold), 0.5 mg/kg (1.5-fold), 1 mg/kg (1.8-fold), and 2 mg/kg (>4-fold). Measurement of intra-tumoral drug levels is in progress. In combination with RT in GBM108, brigimadlin (2 mg/kg x 2 weeks) extended median survival vs. vehicle (66 vs. 50 days; p=0.012) and enhanced median survival following RT (128.5 vs. 82 days; p=0.009). In GBM14, 2 mg/kg brigimadlin (2 mg/kg x 12 weeks) extended median survival (39.5 vs. 32.5 days; p=0.0004) and enhanced survival following RT (97 vs. 81 days; p<0.0001). Conclusions: Efficacy of brigimadlin is dependent on adequate CNS delivery. Studies in subcutaneous PDX demonstrated intra-tumoral total drug levels in the mid-nanomolar range provide meaningful tumor effects in a highly sensitive, MDM2 amplified PDX. In orthotopic PDXs with and without MDM2 amplification, brigimadlin provided further benefit in combination with RT. These data support ongoing clinical evaluation of brigimadlin in GBM.

BACKGROUND: Predictive performance of polygenic risk scores (PRS) varies across populations. To facilitate equitable clinical use, we developed PRS for coronary heart disease (CHD; PRS CHD ) for 5 genetic ancestry groups. METHODS: We derived ancestry-specific and multi-ancestry PRS CHD based on pruning and thresholding (PRS PT ) and ancestry-based continuous shrinkage priors (PRS CSx ) applied to summary statistics from the largest multi-ancestry genome-wide association study meta-analysis for CHD to date, including 1.1 million participants from 5 major genetic ancestry groups. Following training and optimization in the Million Veteran Program, we evaluated the best-performing PRS CHD in 176,988 individuals across 9 diverse cohorts. RESULTS: Multi-ancestry PRS PT and PRS CSx outperformed ancestry-specific PRS PT and PRS CSx across a range of tuning values. Two best-performing multi-ancestry PRS CHD (ie, PRS PTmult and PRS CSxmult ) and 1 ancestry-specific (PRS CSxEUR ) were taken forward for validation. PRS PTmult demonstrated the strongest association with CHD in individuals of South Asian ancestry and European ancestry (odds ratio per 1 SD [95% CI, 2.75 [2.41–3.14], 1.65 [1.59–1.72]), followed by East Asian ancestry (1.56 [1.50–1.61]), Hispanic/Latino ancestry (1.38 [1.24–1.54]), and African ancestry (1.16 [1.11–1.21]). PRS CSxmult showed the strongest associations in South Asian ancestry (2.67 [2.38–3.00]) and European ancestry (1.65 [1.59–1.71]), lower in East Asian ancestry (1.59 [1.54–1.64]), Hispanic/Latino ancestry (1.51 [1.35–1.69]), and the lowest in African ancestry (1.20 [1.15–1.26]). CONCLUSIONS: The use of summary statistics from a large multi-ancestry genome-wide meta-analysis improved the performance of PRS CHD in most ancestry groups compared with single-ancestry methods. Despite the use of one of the largest and most diverse sets of training and validation cohorts to date, improvement of predictive performance was limited in African ancestry. This highlights the need for larger genome-wide association study datasets of underrepresented populations to enhance the performance of PRS CHD .

Abstract Clinical management of IDH wildtype glioblastoma (GBM) and tumefactive multiple sclerosis (tMS) is drastically different. GBM requires maximal safe resection followed by chemoradiation, while tMS outcome is worsened by surgery and radiotherapy. Misdiagnosis of brain lesions may expose patients to unnecessary anxiety, surgery, or radiotherapy. Noninvasive methods to help with accurate diagnosis of tumor and non-tumor etiologies are needed. tMS subjects were identified by the Mayo Clinic Center for Multiple Sclerosis and Autoimmune Neurology (CMSAN). GBM subjects were matched to tMS by age at diagnosis, index MRI date, and sex. Inclusion criteria included one cm minimal lesion size and pre-operative post-contrast T1 and T2 images available for analysis. To identify potential experimental design bias, MRI technology and clinical characteristics were compared across groups. A 3D-DenseNet121 was used to develop a classification model using prespecified parameters: 600 epochs, batch size 16, learning rate 10-3, cross entropy loss, and AdamW optimizer. The stopping rule was defined as three sequential epoch cross entropy loss &lt; 0.02. Five-fold cross validation (CV) was utilized to obtain an unbiased estimate of predictive performance (overfitting) and sensitivity of the 3D-DenseNet121 algorithm was evaluated. 224 subjects (114 GBM, 110 tMS) were analyzed: 40% male, median age 46. After one round of CV, the average unbiased estimate of sensitivity and specificity for predicting GBM from the final classification model was 0.86 and 0.73, respectively. To assess the reproducibility of the deep learning algorithm (3D-DenseNet121), the exact same CV folds were rerun resulting in average sensitivity and specificity of 0.82 and 0.75, respectively. Obtaining unbiased estimates of model performance using internal validation is critical before evaluating a final model on an external dataset. Understanding sensitivity of a deep learning algorithm is important for model validity. Differentiation of GBM and tMS appears reasonable with MRI; external validation is in progress.

Abstract INTRODUCTION Prior to 2021, (classic) glioblastoma (GBM) required absence of IDH mutation and histologic features of hypercellularity, nuclear atypia, microvascular proliferation (MVP), and necrosis. Absence of histologic features can be a result of undersampling, with molecular, radiologic, and clinical features consistent with GBM. Accordingly, WHO 2021 defined molecular glioblastoma as diffuse glioma IDHwt, H3wt, with one or more of MVP, necrosis, TERT promoter mutation, EGFR gene amplification (EGFRamp), and/or +7/−10. We hypothesized that there is clinical and radiographic heterogeneity among tumors defined as molecular GBM. METHODS We identified 121 patients that met the criteria for molecular GBM. 121 Classic GBM, defined as IDHwt with hypercellularity, MVP or necrosis, were matched to molecular GBM cases by age, sex, and diagnosis date. All cases were reviewed for relevant histologic features and molecular features. Examined clinical variables included: first symptom, time from first symptom to date of pathological diagnosis, date of first MRI, tumor location, radiographic features of enhancement, diffusion or mass effect, time to progression and overall survival. Clinical, radiologic, and molecular features were compared between classic and molecular GBM. RESULTS Currently,160 cases have been reviewed. Preliminary results show that there is a longer time from first symptom to diagnosis for molecular GBM. Molecular GBM is significantly more likely to be discovered coincidentally, and without enhancement or radiographic necrosis. Classic GBM is significantly more likely to present with focal neurologic deficit as first symptom, regional mass effect on index imaging and have monosomy 10. CONCLUSION While pathologic undersampling accounts for a significant proportion of molecular GBM, our data suggest that there are significant clinical, radiographic, and outcome differences between classical and molecular GBM. Final analysis of the entire cohort will help better define this clinical spectrum.

Introduction. We assessed the effects of a polygenic risk score (PRS), monogenic variants (Mono + ), and family history (FamHx) on the risk of coronary heart disease (CHD) and whether incorporating a PRS into the Pooled Cohort Equations (PCE) improves risk prediction. Method. In 199,997 UK Biobank participants, we calculated a PRS CHD (metaGRS, based on ~1.6 million single nucleotide variants), identified individuals with monogenic etiology (presence of a pathogenic variant in LDLR , APOB , or PCSK9 ), and ascertained family history of CHD (CHD in a first-degree relative). Odds of CHD (myocardial infarction/ coronary revascularization/ cardiovascular death) associated with PRS CHD , Mono + , and FamHx at age 55 were computed using logistic regression analyses adjusted for age, sex, and first 4 principal components of ancestry. We combined PRS CHD and PCE into an integrated score (IS) and assessed net reclassification using an actionable risk threshold of 7.5%. Results. The odds ratio (OR) for CHD in the top 5 th percentile of PRS CHD (PRS P95 ), Mono + , and FamHx were 3.37 (2.98 - 3.81), 2.68 (1.67 - 4.07), and 1.61 (1.47 - 1.76), respectively (all p<0.001). These effects were independent and similar in different subgroups. ( Figure 1-A ) Compared to PRS Q2-4 Mono - (3 middle quintiles of PRS CHD and no monogenic variants), the OR for CHD ranged from 1.73 (0.28 - 5.48, p = 0.446) in PRS Q1 Mono + to 5.10 (2.14 - 10.24, p <0.001) in PRS Q5 Mono + . A similar pattern was seen for other combinations, indicating additive effects ( Figure 1-B ). Incorporating PRS CHD into PCE resulted in a net reclassification improvement of 4.56%. Compared to those with high PCE and low IS, the hazard ratio for CHD in those with low PCE and high IS was 1.48 (1.34 - 1.64, p <0.001). Conclusion. The CHD risk associated with PRS CHD , monogenic variants, and family history is independent and additive. Combining PRS CHD and PCE improved risk prediction, that could improve further by adding monogenic variants and family history.

Abstract Missing MRI sequences are common in radiology due to differences in protocols, time constraints, and image artifacts. Missing sequences are a nuisance for subsequent algorithms, such as automated segmentation algorithms requiring multiple MRI sequences (e.g., HD-GLIO, which requires T2, T1, post-contrast T1 (T1c), and FLAIR). Accurate tumor segmentation is a fundamental step in extracting radiomics analysis to use, for example, to predict the IDH status of glioma or their overall survival. We validated a conditional (pix2pix) and super-resolution GAN (SOUP-GAN) for imputing MRI sequences on 285 adult glioma with complete MRI sequences available. We simulated missing T1 sequences and subsequently imputed them from T1c using pix2pix and from sagittal T1 using SOUP-GAN. Using HD-GLIO, we obtained the true segmentations from the complete MRI sequences. HD-GLIO was also used after replacing the original T1 sequence with the one imputed with pix2pix and SOUP-GAN. Thus, there were three sets of segmentations for each subject. We compared T2 and T1c radiomics features extracted with Pyradiomics using segmentations obtained with complete data with features extracted from segmentation obtained with the imputed T1 sequences using intraclass correlation coefficient (ICC) and Bland-Altman (BA) analysis. Median (interquartile range) ICC for shape features extracted from T2 and T1c were 0.82(0.76-0.92) using pix2pix and 0.89(0.86-0.94) using SOUP-GAN. Median ICC for first-order-statistics features were 0.90(0.83-0.93) for T2 and 0.90(0.88-0.92) for T1c using pix2pix, and 0.61(0.22-0.82) for T2 and 0.42(0.34-0.57) for T1c using SOUP-GAN. Median ICC for texture features were 0.86(0.82-0.87) for T2 and 0.87(0.79-0.90) for T1c using pix2pix, and 0.69(0.47-0.81) for T2 and 0.76(0.65-0.83) for T1c using SOUP-GAN. We show that T2 and T1c shape features (e.g., volumetric analysis) can be adequately extracted from segmentations obtained with an algorithm requiring multiple MRI sequences using T1 sequences imputed with GANs; however, the effect of imputation differed for first-order-statistics and texture features.

Establishing causal links between inherited polymorphisms and cancer risk is challenging. Here, we focus on the single-nucleotide polymorphism rs55705857, which confers a sixfold greater risk of isocitrate dehydrogenase ( IDH) –mutant low-grade glioma (LGG). We reveal that rs55705857 itself is the causal variant and is associated with molecular pathways that drive LGG. Mechanistically, we show that rs55705857 resides within a brain-specific enhancer, where the risk allele disrupts OCT2/4 binding, allowing increased interaction with the Myc promoter and increased Myc expression. Mutating the orthologous mouse rs55705857 locus accelerated tumor development in an Idh1 R132H -driven LGG mouse model from 472 to 172 days and increased penetrance from 30% to 75%. Our work reveals mechanisms of the heritable predisposition to lethal glioma in ~40% of LGG patients.