Asmann, Yan W.

Abstract Explainable artificial intelligence aims to interpret how machine learning models make decisions, and many model explainers have been developed in the computer vision field. However, understanding of the applicability of these model explainers to biological data is still lacking. In this study, we comprehensively evaluated multiple explainers by interpreting pre-trained models for predicting tissue types from transcriptomic data and by identifying the top contributing genes from each sample with the greatest impacts on model prediction. To improve the reproducibility and interpretability of results generated by model explainers, we proposed a series of optimization strategies for each explainer on two different model architectures of multilayer perceptron (MLP) and convolutional neural network (CNN). We observed three groups of explainer and model architecture combinations with high reproducibility. Group II, which contains three model explainers on aggregated MLP models, identified top contributing genes in different tissues that exhibited tissue-specific manifestation and were potential cancer biomarkers. In summary, our work provides novel insights and guidance for exploring biological mechanisms using explainable machine learning models.

Abstract Background In the United States, cancer disproportionately impacts Black and African American individuals. Identifying genetic factors underlying cancer disparities has been an important research focus and requires data that are equitable in both quantity and quality across racial groups. It is widely recognized that DNA databases quantitatively underrepresent minorities. However, the differences in data quality between racial groups have not been well studied. Methods We compared the qualities of germline and tumor exomes between ancestrally African and European patients in The Cancer Genome Atlas of 7 cancers with at least 50 self-reported Black patients in the context of sequencing depth, tumor purity, and qualities of germline variants and somatic mutations. Results Germline and tumor exomes from ancestrally African patients were sequenced at statistically significantly lower depth in 6 out of the 7 cancers. For 3 cancers, most ancestrally European exomes were sequenced in early sample batches at higher depth, whereas ancestrally African exomes were concentrated in later batches and sequenced at much lower depth. For the other 3 cancers, the reasons of lower sequencing coverage of ancestrally African exomes remain unknown. Furthermore, even when the sequencing depths were comparable, African exomes had disproportionally higher percentages of positions with insufficient coverage, likely because of the known European bias in the human reference genome that impacted exome capture kit design. Conclusions Overall and positional lower sequencing depths of ancestrally African exomes in The Cancer Genome Atlas led to underdetection and lower quality of variants, highlighting the need to consider epidemiological factors for future genomics studies.

Abstract Frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP) is a complex heterogeneous neurodegenerative disorder for which mechanisms are poorly understood. To explore transcriptional changes underlying FTLD-TDP, we performed RNA-sequencing on 66 genetically unexplained FTLD-TDP patients, 24 FTLD-TDP patients with GRN mutations and 24 control participants. Using principal component analysis, hierarchical clustering, differential expression and coexpression network analyses, we showed that GRN mutation carriers and FTLD-TDP-A patients without a known mutation shared a common transcriptional signature that is independent of GRN loss-of-function. After combining both groups, differential expression as compared to the control group and coexpression analyses revealed alteration of processes related to immune response, synaptic transmission, RNA metabolism, angiogenesis and vesicle-mediated transport. Deconvolution of the data highlighted strong cellular alterations that were similar in FTLD-TDP-A and GRN mutation carriers with NSF as a potentially important player in both groups. We propose several potentially druggable pathways such as the GABAergic, GDNF and sphingolipid pathways. Our findings underline new disease mechanisms and strongly suggest that affected pathways in GRN mutation carriers extend beyond GRN and contribute to genetically unexplained forms of FTLD-TDP-A.

Abstract TMB is used to guide PD-1-directed immunotherapy in solid tumor Oncology. However, it has not been systematically studied in AML, where the focus has been on cytogenetic risk and individual driver gene mutations (GM's). TMB contribution to AML epidemiology is also uncertain. We therefore studied its association with epidemiologic risk factors; driver GM's and somatic mutations (SM's) in AML risk genes which we recently demonstrated (ABCB1; CYP1A1; CYP2B6; EPHX1; ERCC1,2,& 5; MEFV; MTRR; and TERT); clinical and cytogenetic features; and outcome after therapy in the MCAEC, a highly annotated clinical epidemiology cohort of consecutive AML pts [Finn, Cancer Epidemiol 39:1084, 2015]. Methods: We obtained somatic leukemia DNA from remnant diagnostic cytogenetic pellets in 98 MCAEC patients (pts), as previously described [Foran, Blood (2017) 130:570a]. Whole exome sequencing (WES) was performed at depth of ~100 million paired end 100bp reads using Agilent SureSelectXT Human All Exon V5 + UTRs target enrichment kit. Reads were mapped to human genome build hg19 using BWA-MEM. Single nucleotide variants (SNVs) and small INDELs were identified using Mayo Clinic (MC) analytic pipeline GenomeGPS 4.0.1 following Broad GATK variant discovery best practices of alignment, realignment and recalibration, and multi-sample joint genotyping; and filtered using both germline whole exome and genome sequencing of ~1200 MC Biobank samples and public germline variant databases of 1000 genome project, 6500 individuals in exome sequencing project, and HapMap phase 3. Remaining variants were annotated using ANNOVAR, and functional variants of non-synonymous, truncating, frame-shift, and splice-sites were used in the statistical association analyses. TMB was defined as the number of functional mutations per Mb of coding region, heterozygous or homozygous. TMB associations with epidemiologic risk, clinical characteristics, and SM's in AML risk genes (listed above) or driver GM's (occurring in 5 or more pts: ASXL1, BCOR, CEBPA, DNMT3A, FLT3, IDH2, KRAS, MLL2-5, NF1, NPM1, NRAS, PHF6, RUNX1, SF3B1, STAG2, TET2, TP53, U2AF1) were evaluated using linear regression models; a rank transformation of TMB was utilized due to its skewed distribution. Multivariable analysis (MVA) models were adjusted for variables with p-value <0.05 in unadjusted analysis. Regression coefficients ("β") were estimated, interpreted as change in mean TMB rank for specified variables or characteristic. Associations with outcomes were examined using logistic regression or Cox proportional hazards regression models. All statistical tests were two-sided (version 3.6.2; R Foundation for Statistical Computing, Vienna, Austria). Results Median age at AML diagnosis was 70 yrs (Range: 19-94 yrs), and 67 pts were male. Cytogenetic risk group was favorable (7%), intermediate-normal (46%) or abnormal (20%), and poor risk (27%). 40/61 pts (66%) achieved complete remission (CR). With a median follow-up of 8.0 months (Range: 0.1 - 186), 80 pts (82%) died and 20 (20%) underwent allogeneic transplantation (AlloBMT). Median TMB was 18.2 (Range: 15.0-70.1). In MVA, significant associations with increased TMB were observed in pts with history of prior immunosuppressive therapy or solid organ transplantation (β=19.48, P=0.015), and with FLT3 (β=21.12, P=0.015), MLL2 (β=20.91, P=0.001), and MLL3 (β=11.31, P=0.031) GM's. A borderline association was observed for U2AF1 (β=16.14, P=0.057). TMB was also associated with SM's in TERT (β=25.13, P=0.028); a borderline association with SM's in ABCB1 was not confirmed in MVA (β=-17.98, P=0.069). Additionally, cytogenetic risk group was associated with TMB in MVA (P=0.005), being highest in intermediate-normal and lowest in poor risk groups. Body Mass Index was inversely associated with TMB (unadjusted β=-16.99, P=0.005), but unconfirmed in MVA (β=-8.29, P=0.12). There was no association with CR (OR=0.93, P=0.46), use of (HR=0.96, P=0.64) or relapse risk (HR=1.00, P=0.98) following AlloBMT, or survival (HR=0.97, P=0.56) (Figure). Conclusions Measurement of TMB is feasible in this AML epidemiologic cohort, and we observed important associations with AML risk factors, risk gene SM's, cytogenetic risk group, and driver GM's. We acknowledge the relatively small sample size and possibility of type II error, and therefore these observations require validation in a large prospective cohort which is planned. Figure 1 Figure 1. Disclosures Foran: OncLive: Honoraria; certara: Honoraria; actinium: Research Funding; boehringer ingelheim: Research Funding; novartis: Honoraria; abbvie: Research Funding; servier: Honoraria; taiho: Honoraria; pfizer: Honoraria; revolution medicine: Honoraria; gamida: Honoraria; takeda: Research Funding; sanofi aventis: Honoraria; trillium: Research Funding; syros: Honoraria; aptose: Research Funding; bms: Honoraria; kura: Research Funding; h3bioscience: Research Funding; aprea: Research Funding; sellas: Research Funding; stemline: Research Funding. Murthy: CRISPR Therapeutics: Research Funding. Finn: Jazz: Consultancy, Speakers Bureau; BMS: Consultancy, Speakers Bureau; BeiGene: Consultancy, Speakers Bureau; ADC Therapeutics: Consultancy, Speakers Bureau. Badar: Pfizer Hematology-Oncology: Membership on an entity's Board of Directors or advisory committees. Cerhan: Celgene/BMS: Other: Connect Lymphoma Scientific Steering Committee, Research Funding; Genentech: Research Funding; NanoString: Research Funding; Regeneron Genetics Center: Other: Research Collaboration.

Abstract Background: Chimeric antigen receptor T cell therapy (CAR-T) is a recent therapeutic advance for the treatment of myeloma. Among the FDA approved and investigational CAR-T, primary refractory disease post CAR-T infusion is uncommon, however most patients eventually relapse. To gain insight to the myeloma cell and immune cell characteristics associated with early relapse (PD) versus durable response (DR), we examined myeloma cells and immune cells in both the bone marrow and blood by single cell RNA-seq. Method: Bone marrow aspirate and blood samples were collected prior to BCMA targeted CAR-T therapy. CD138 + and CD138 neg cells from bone marrow (BM) and peripheral blood mononuclear cells (PBMNC) were collected for 5' based scRNA-seq (10X Genomics). CD138 + cells were clustered by Harmony integration. Top expressing genes in unique clusters for PD and DR were analyzed by Ingenuity Pathway Analyses. To profile CD138 neg BM cells and PBMNC from myeloma patients, the scRNA-seq dataset was mapped to the human PBMC reference using the Seurat v4 reference-guiding approach. Differential gene expression between responders and non-responders was performed using the FindMarkers function for each of the immune cell clusters in the BM and PB. Results: BM and PB samples from 15 patients were analyzed. Among these, 5 patients had relapsed disease more than 1-year post CAR-T infusion (Durable Response, DR), and 10 patients had relapse within 1 year (early relapse, PD). Among these, CD138+ cells were available from 3 DR and 4 PD for analysis. Multimodal expression of BCMA were seen among the CD138+ cells across all samples. Using the mode of the highest expression level as a cut-off, patients with DR had a trend toward having higher fraction of BCMA high CD138+ cells compared to pts with PD. Clustering of all CD138+ cells by Harmony integration identified 12 distinct clusters, among which two clusters were unique to pts in DR, and one was unique to patients in PD. Ingenuity pathway analysis identified the top marker genes in these 3 clusters to be enriched in pathways for IL-15 signaling, BCR signaling, and primary immunodeficiency signaling. (Figure 1A-E.) Differential gene expression for immune cell clusters from CD138 neg BM cells and PBMNC were compared between pts in PD and DR. Interestingly, different patterns were seen in immune cells in PB and BM (Figure 1G). For example, compared to pts in PD, those with DR had decreased signaling in CD16 PB monocytes for interferon signaling and differentiation to macrophages, whereas the CD16 monocytes in the BM in pts in DR had increased expression of mitochondrial genes for anti-oxidant stress, polarization to M1 macrophage and anti-apoptosis. Similarly, pts in DR had CD8 Tcm in the BM with increased expression of genes involved in cell adhesion and anti-apoptosis, and CD4 Tcm in the BM with increased expression of genes in protein metabolism. Conclusion: scRNAseq analysis of myeloma cells suggest different tumor transcriptome profile may be identified in myeloma cells that could relapse early after CAR-T therapy. In addition, host immune profile both in the BM microenvironment and in systemic PB circulation could be associated with durable clinical response. Figure 1 Figure 1. Disclosures Kapoor: Pharmacyclics: Consultancy; Glaxo SmithKline: Research Funding; Amgen: Research Funding; Sanofi: Research Funding; AbbVie: Research Funding; Ichnos Sciences: Research Funding; Sanofi: Consultancy; Regeneron Pharmaceuticals: Research Funding; Takeda: Research Funding; Karyopharm: Research Funding; Karyopharm: Consultancy; Cellectar: Consultancy; BeiGene: Consultancy. Dingli: GSK: Consultancy; Janssen: Consultancy; Sanofi: Consultancy; Novartis: Research Funding; Apellis: Consultancy; Alexion: Consultancy. Kumar: Bluebird Bio: Consultancy; Janssen: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Takeda: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; KITE: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Tenebio: Research Funding; Novartis: Research Funding; BMS: Consultancy, Research Funding; Celgene: Membership on an entity's Board of Directors or advisory committees, Research Funding; Roche-Genentech: Consultancy, Research Funding; Astra-Zeneca: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Abbvie: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Antengene: Consultancy, Honoraria; Carsgen: Research Funding; Oncopeptides: Consultancy; Beigene: Consultancy; Amgen: Consultancy, Research Funding; Merck: Research Funding; Adaptive: Membership on an entity's Board of Directors or advisory committees, Research Funding; Sanofi: Research Funding. Lin: Merck: Research Funding; Kite, a Gilead Company: Consultancy, Research Funding; Takeda: Research Funding; Bluebird Bio: Consultancy, Research Funding; Celgene: Consultancy, Research Funding; Sorrento: Consultancy; Juno: Consultancy; Legend: Consultancy; Gamida Cell: Consultancy; Vineti: Consultancy; Novartis: Consultancy; Janssen: Consultancy, Research Funding.

Abstract Purpose: Multiple myeloma (MM) is a plasma cell (PC) malignancy with an increasing incidence in the United States. Fluorescence in situ hybridization (FISH) is currently the gold-standard diagnostic assay to detect recurrent genomic abnormalities of prognostic and therapeutic significance in MM. Hyperdiploid MM, a standard risk abnormality, is characterized by gains of odd-numbered chromosomes including chromosome 15. While gains of chromosome 15 have been reported in approximately 53% of standard risk MM patients (Chretien, Blood, 2015), losses of chromosome 15 are rarely reported. Enumeration of chromosome 15 centromere is detected using the alpha satellite FISH probe. We previously discovered ~9% of cases had false-negative chromosome 15 FISH results when compared to Mate Pair Sequencing (MPseq) (Smadbeck, BCJ, 2019). To our knowledge, no previous studies have reported a reduced performance of the alpha satellite D15Z4 FISH probe in PC FISH studies. A false-negative result for D15Z4 could result in an underreporting of trisomy 15 and overreporting of monosomy 15. Here, we investigated the incidence of false-negative D15Z4 FISH results by examining the presence of monosomy 15 from patients with a monoclonal gammopathy. Methods: We identified 1231 samples from patients with a monoclonal gammopathy who had undergone uniform testing to identify MM-specific cytogenetic abnormalities. DNA from bone marrow samples was genotyped on the Precision Medicine Research Array and biogeographical ancestry was assessed using the Geographic Population Structure Origins tool (Elhaik, Nat Commun, 2014; Home DNA 2016). FISH of immunoglobulin (cIg)-stained PCs was performed as described (Baughn, BCJ, 2018). Chromosome 15 centromere probe (Vysis CEP 15 D15Z4, Abbott Molecular, Abbott Park, IL) was used for chromosome 15 enumeration. Samples were divided into two groups: Cases with or without evidence of monosomy 15 by FISH. Mann-Whitney U test was applied to compare difference between groups for data deviating from normal distribution. A Chi-squared test evaluating the differences across these two groups was used. Results: We identified monosomy 15 in 31 cases (2.5% of cohort). Monosomy 15 was observed in nearly 100% of both the PC (cIg+) and non-PC (cIg-) populations, by contrast to the other recurrent PC-specific genomic abnormalities which were restricted to only the PC population. All samples with a banded analysis revealed most metaphases had two normal chromosome 15s demonstrating that the D15Z4 FISH probe failed to correctly enumerate the number of chromosome 15 centromere signals. Of the samples with a monosomy 15 result, the median African ancestry was 76.2% (95% percentiles: 77.5%-89.0%), while the median African ancestry of the non-monosomy 15 cohort was 2.3% (95% percentiles: 0%-87.7%) (). Although not all patients with >50% African ancestry had evidence of monosomy 15, 26/31 (83.9%) of patients with monosomy 15 had >50% African ancestry. Accordingly, patients with African ancestry (similarity ≥ 0.70) had 8.02-fold (95% confidence interval: 3.73-17.25, P = 9.92 X10 -8) increased possibility to have monosomy 15. Of cases with monosomy 15, 22 (71.0%) had an IGH rearrangement with the most prevalent IGH partner being CCND1. The incidence of t(11;14) was greater in the monosomy 15 group (13/31:41.9%), in contrast to the non-monosomy 15 group (330/1200=27.5%) (P = 0.12). Conclusions: We report a polymorphic variant of the chromosome 15 centromere resulting in a false-negative FISH result using the D15Z4 FISH probe. Surprisingly, the false-negative FISH result was not uniformly distributed among patients and was highly enriched in individuals of African ancestry. This may be due to a variation of the chromosome 15 alpha satellite in some individuals that prevents the D15Z4 FISH probe from sufficiently binding (O'Keefe, Genome Research, 2000). Although this study focused on identification of monosomy 15, the incidence of a false-negative result is likely greater than we have reported in this study since we did not evaluate cases that had a true trisomy 15 that was reported as normal chromosome 15 enumeration by FISH. This study highlights a critical need to ensure diagnostic reagents have equal performance for all patients independent of their ancestry. Evaluation of genomic events using unbiased approaches such as whole genome sequencing may circumvent some of these limitations. Disclosures Elhaik: DDC: Consultancy. Baird: DDC: Current Employment. Kumar: Antengene: Consultancy, Honoraria; Bluebird Bio: Consultancy; Beigene: Consultancy; BMS: Consultancy, Research Funding; Tenebio: Research Funding; Takeda: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Abbvie: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Amgen: Consultancy, Research Funding; Novartis: Research Funding; Astra-Zeneca: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Merck: Research Funding; KITE: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Roche-Genentech: Consultancy, Research Funding; Janssen: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Oncopeptides: Consultancy; Carsgen: Research Funding; Celgene: Membership on an entity's Board of Directors or advisory committees, Research Funding; Adaptive: Membership on an entity's Board of Directors or advisory committees, Research Funding; Sanofi: Research Funding.

A rare APOE variant APOE3 -Jac reduces self-aggregation and promotes healthy brain aging.

PLEXIND1 is upregulated in several cancers, including pancreatic ductal adenocarcinoma (PDAC). It is an established mediator of semaphorin signaling, and neuropilins are its known coreceptors. Herein, we report data to support the proposal that PLEXIND1 acts as a transforming growth factor beta (TGFβ) coreceptor, modulating cell growth through SMAD3 signaling. Our findings demonstrate that PLEXIND1 plays a pro-tumorigenic role in PDAC cells with oncogenic KRAS (KRASmut). We show in KRASmut PDAC cell lines (PANC-1, AsPC-1,4535) PLEXIND1 downregulation results in decreased cell viability (in vitro) and reduced tumor growth (in vivo). Conversely, PLEXIND1 acts as a tumor suppressor in the PDAC cell line (BxPC-3) with wild-type KRAS (KRASwt), as its reduced expression results in higher cell viability (in-vitro) and tumor growth (in vivo). Additionally, we demonstrate that PLEXIND1-mediated interactions can be selectively disrupted using a peptide based on its C-terminal sequence (a PDZ domain-binding motif), an outcome that may possess significant therapeutic implications. To our knowledge, this is the first report showing that (1) PLEXIND1 acts as a TGFβ coreceptor and mediates SMAD3 signaling, and (2) differential roles of PLEXIND1 in PDAC cell lines correlate with KRASmut and KRASwt status.

Abstract Purpose: Structural variants (SV) of the MYC gene region are common in multiple myeloma and influence disease progression. However, the prognostic significance of different MYC SVs in multiple myeloma has not been clearly established. Experimental Design: We conducted a retrospective study of multiple myeloma comparing MYC SV subtypes identified by next-generation sequencing (NGS) and FISH to MYC expression and disease survival using 140 cases from Mayo Clinic and 658 cases from the MMRF CoMMpass study. Results: MYC SVs were found in 41% of cases and were classified into nine subtypes. A correlation between the presence of a MYC SV and increased MYC expression was identified. Among the nine MYC subtypes, the non-immunoglobulin (non-Ig) insertion subtype was independently associated with improved outcomes, while the Ig insertion subtype, specifically involving the IgL gene partner, was independently associated with poorer outcomes compared with other MYC SV subtypes. Although the FISH methodology failed to detect approximately 70% of all MYC SVs, those detected by FISH were associated with elevated MYC gene expression and poor outcomes suggesting a different pathogenic role for FISH-detected MYC subtypes compared with other MYC subtypes. Conclusions: Understanding the impact of different MYC SVs on disease outcome is necessary for the reliable interpretation of MYC SVs in multiple myeloma. NGS approaches should be considered as a replacement technique for a more comprehensive evaluation of the multiple myeloma clone.