Whole-genome analysis identifies novel drivers and high-risk double-hit events in relapsed/refractory myeloma

Large-scale analyses of genomic data from patients with newly diagnosed multiple myeloma (ndMM) have been undertaken, however, large-scale analysis of relapsed/refractory MM (rrMM) has not been performed. We hypothesize that somatic variants chronicle the therapeutic exposures and clonal structure of myeloma from ndMM to rrMM stages. We generated whole-genome sequencing (WGS) data from 418 tumors (386 patients) derived from 6 rrMM clinical trials and compared them with WGS from 198 unrelated patients with ndMM in a population-based case-control fashion. We identified significantly enriched events at the rrMM stage, including drivers (DUOX2, EZH2, TP53), biallelic inactivation (TP53), noncoding mutations in bona fide drivers (TP53BP1, BLM), copy number aberrations (CNAs; 1qGain, 17pLOH), and double-hit events (Amp1q-ISS3, 1qGain-17p loss-of-heterozygosity). Mutational signature analysis identified a subclonal defective mismatch repair signature enriched in rrMM and highly active in high mutation burden tumors, a likely feature of therapy-associated expanding subclones. Further analysis focused on the association of genomic aberrations enriched at different stages of resistance to immunomodulatory agent (IMiD)-based therapy. This analysis revealed that TP53, DUOX2, 1qGain, and 17p loss-of-heterozygosity increased in prevalence from ndMM to lenalidomide resistant (LENR) to pomalidomide resistant (POMR) stages, whereas enrichment of MAML3 along with immunoglobulin lambda (IGL) and MYC translocations distinguished POM from the LEN subgroup. Genomic drivers associated with rrMM are those that confer clonal selective advantage under therapeutic pressure. Their role in therapy evasion should be further evaluated in longitudinal patient samples, to confirm these associations with the evolution of clinical resistance and to identify molecular subsets of rrMM for the development of targeted therapies.

Loss of COP9 signalosome genes at 2q37 is associated with IMiD resistance in multiple myeloma

The acquisition of a multidrug refractory state is a major cause of mortality in myeloma. Myeloma drugs that target the cereblon (CRBN) protein include widely used immunomodulatory drugs (IMiDs), and newer CRBN E3 ligase modulator drugs (CELMoDs), in clinical trials. CRBN genetic disruption causes resistance and poor outcomes with IMiDs. Here, we investigate alternative genomic associations of IMiD resistance, using large whole-genome sequencing patient datasets (n = 522 cases) at newly diagnosed, lenalidomide (LEN)-refractory and lenalidomide-then-pomalidomide (LEN-then-POM)-refractory timepoints. Selecting gene targets reproducibly identified by published CRISPR/shRNA IMiD resistance screens, we found little evidence of genetic disruption by mutation associated with IMiD resistance. However, we identified a chromosome region, 2q37, containing COP9 signalosome members COPS7B and COPS8, copy loss of which significantly enriches between newly diagnosed (incidence 5.5%), LEN-refractory (10.0%), and LEN-then-POM-refractory states (16.4%), and may adversely affect outcomes when clonal fraction is high. In a separate dataset (50 patients) with sequential samples taken throughout treatment, we identified acquisition of 2q37 loss in 16% cases with IMiD exposure, but none in cases without IMiD exposure. The COP9 signalosome is essential for maintenance of the CUL4-DDB1-CRBN E3 ubiquitin ligase. This region may represent a novel marker of IMiD resistance with clinical utility.

Interactome of Aiolos/Ikaros Reveals Combination Rationale of Cereblon Modulators with HDAC Inhibitors in DLBCL

PURPOSE

Cereblon (CRBN), a substrate receptor of the E3 ubiquitin ligase complex CRL4CRBN, is the target of the small molecules lenalidomide and avadomide. Upon binding of the drugs, Aiolos and Ikaros are recruited to the E3 ligase, ubiquitylated, and subsequently degraded. In diffuse large B-cell lymphoma (DLBCL) cells, Aiolos and Ikaros are direct transcriptional repressors of interferon-stimulated genes (ISG) and degradation of these substrates results in increased ISG protein levels resulting in decreased proliferation and apoptosis. Herein, we aimed to uncover the mechanism(s) Aiolos and Ikaros use to repress ISG transcription and provide a mechanistic rationale for a combination strategy to enhance cell autonomous activities of CRBN modulators (CELMoD).

EXPERIMENTAL DESIGN

We conducted paired RNA sequencing with histone modification and Aiolos/Ikaros chromatin immunoprecipitation sequencing to identify genes regulated by these transcription factors and to elucidate correlations to drug sensitivity. We confirmed Aiolos/Ikaros mediated transcriptional complex formation in DLBCL patient samples including those treated with avadomide.

RESULTS

In DLBCL, the repression of ISG transcription is accomplished in part through recruitment of large transcriptional complexes such as the nucleosome remodeling and deacetylase, which modify the chromatin landscape of these promoters. A rational combination approach of avadomide with a specific histone deacetylase inhibitor leads to a significant increase in ISG transcription compared with either single agent, and synergistic antiproliferative activity in DLBCL cell lines.

CONCLUSIONS

Our results provide a novel role for lineage factors Aiolos and Ikaros in DLBCL as well as further insight into the mechanism(s) of Aiolos and Ikaros-mediated transcriptional repression and unique therapeutic combination strategies.

Multiple Myeloma Patient Tumors With High Levels of Cereblon Exon-10 Deletion Splice Variant Upregulate Clinically Targetable Pro-Inflammatory Cytokine Pathways

Immunomodulatory drugs (IMiDs), including lenalidomide and pomalidomide, are used in the routine treatment for multiple myeloma (MM) patients. Cereblon (CRBN) is the direct molecular target of IMiDs. While CRBN is not an essential gene for MM cell proliferation, the frequency of CRBN genetic aberrations, including mutation, copy number loss, and exon-10 (which includes a portion of the IMiD-binding domain) splicing, have been reported to incrementally increase in later-line patients. CRBN exon-10 splicing has also been shown to be associated with decreased progression-free survival in both newly diagnosed and relapsed refractory MM patients. Although we did not find significant general splicing defects among patients with CRBN exon-10 splice variant when compared to those expressing the full-length transcript, we identified upregulated TNFA signaling via NFKB, inflammatory response, and IL-10 signaling pathways in patients with exon-10 splice variant across various data sets—all potentially promoting tumor growth via chronic growth signals. We examined master regulators that mediate transcriptional programs in CRBN exon-10 splice variant patients and identified BATF, EZH2, and IKZF1 as the key candidates across the four data sets. Upregulated downstream targets of BATF, EZH2, and IKZF1 are components of TNFA signaling via NFKB, IL2/STAT5 signaling pathways, and IFNG response pathways. Previously, BATF-mediated transcriptional regulation was associated with venetoclax sensitivity in MM. Interestingly, we found that an EZH2 sensitivity gene expression signature also correlated with high BATF or venetoclax sensitivity scores in these tumors. Together, these data provide a rationale for investigating EZH2 inhibitors or venetoclax in combination with the next generation CRBN-targeting agents, such as CELMoDs, for patients overexpressing the CRBN exon-10 splice variant.

Integrative multi-omics identifies high risk multiple myeloma subgroup associated with significant DNA loss and dysregulated DNA repair and cell cycle pathways

Background

Despite significant therapeutic advances in improving lives of multiple myeloma (MM) patients, it remains mostly incurable, with patients ultimately becoming refractory to therapies. MM is a genetically heterogeneous disease and therapeutic resistance is driven by a complex interplay of disease pathobiology and mechanisms of drug resistance. We applied a multi-omics strategy using tumor-derived gene expression, single nucleotide variant, copy number variant, and structural variant profiles to investigate molecular subgroups in 514 newly diagnosed MM (NDMM) samples and identified 12 molecularly defined MM subgroups (MDMS1-12) with distinct genomic and transcriptomic features.

Results

Our integrative approach let us identify NDMM subgroups with transversal profiles to previously described ones, based on single data types, which shows the impact of this approach for disease stratification. One key novel subgroup is our MDMS8, associated with poor clinical outcome [median overall survival, 38 months (global log-rank p-value < 1 × 10⁻⁶)], which uniquely presents a broad genomic loss (> 9% of entire genome, t-test p value < 1e−5) driving dysregulation of various transcriptional programs affecting DNA repair and cell cycle/mitotic processes. This subgroup was validated on multiple independent datasets, and a master regulator analyses identified transcription factors controlling MDMS8 transcriptomic profile, including CKS1B and PRKDC among others, which are regulators of the DNA repair and cell cycle pathways.

Conclusion

Using multi-omics unsupervised clustering we were able to discover a new high-risk multiple myeloma patient segment. This high-risk group presents diverse previously known genetic markers, but also a new characteristic defined by accumulation of genomic loss which seems to drive transcriptional dysregulation of cell cycle, DNA repair and DNA damage. Finally, our work identified various master regulators, including E2F2 and CKS1B as the genes controlling these key biological pathways.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12920-021-01140-5.

Abstract PO-008: Multi-omic Profiling of primary pancreatic adenocarcinomas obtained from the APACT adjuvant trial of nab-paclitaxel + gemcitabine vs gemcitabine

Pancreatic ductal adenocarcinoma (PDAC) remains a difficult disease to treat, with few therapies available that target specific patient subgroups. Translational studies in pancreatic cancer can be technically challenging due to biopsy specimen characteristics including low tumor cellularity and dense fibrotic stroma. Resected primary pancreatic tumors obtained during the phase 3 APACT clinical trial (NCT01964430 adjuvant nab-paclitaxel + gemcitabine versus gemcitabine monotherapy n = 866) are a unique resource for characterizing molecular and immune subtypes among PDAC tumors and their association with treatment outcomes in the adjuvant chemotherapy setting. Tumor-infiltrating immune cells were assessed in serial sections of 453 tumors by dual-chromogen immuno-histochemical (IHC) staining (CD4 CD8 CD20 CD163 CMAF CD56 FoxP3 PD-1 PD-L1). Image alignment, segmentation, and spatial localization of stained cells relative to a pan-cytokeratin staining-based tumor epithelial mask was performed using a commercial analysis platform. Expression profiles were obtained for 515 macrodissected tumor biopsy regions by RNA-seq. Transcriptional subtypes were assigned based on schema previously reported by Moffit and Bailey, and molecular pathway correlates were characterized using gene set enrichment analysis. Based on immunochemical staining, higher intratumoral CD8+ or lower CD163+ cell densities were associated with modestly longer disease-free or overall survival in patients treated with nab-paclitaxel plus gemcitabine. The combination of both high CD8+ and low CD163+ cell density was notably associated with longer overall survival compared to other subjects treated with the combination regimen (mOS&gt;55months versus 36 months HR=0.46 p=&lt;0.01) Transcriptional subtyping of tumors, based on schema such as those previously reported by Moffit et al or Bailey et al, identified groups defined by anticipated signatures. Classical/progenitor subtypes differentially expressed genes involved in oxidative respiration and endodermal cell fate, while basal/squamous subtypes showed high expression of signatures for hypoxia, mesenchymal transformation, and TGFb signaling. Longer overall survival in subjects treated with nab-paclitaxel plus gemcitabine as compared to gemcitabine monotherapy was seen principally in subjects with tumors of Moffit classical (HR=0.59 P=0.01) and Bailey progenitor (HR=0.42 P&lt;0.01) subtypes. The observations highlighted provide insight into primary PDAC tumor characteristics that are associated with treatment-specific outcomes to nab-paclitaxel based adjuvant chemotherapy. Future work will combine data described here with mutational and genetic studies in progress as well as with other public, proprietary, or commercial data sources. In combination, we anticipate that these data may be useful in identifying patient subsets and molecular mechanisms that might be targeted by novel combination therapies for improved treatment of pancreatic cancer.

Citation Format: Thomas Lila, Andrew Biankin, Andrew Browne, David J. Reiss, Brian Lu, Daniel Pierce, Alexander Ratushny, Kao-tai Tsai, Sneh Lata, Sitharthan Kamalakaran, Tomas Babak, Brian Fox, Sam Danziger, Konstantinos Mavrommatis, Matthew W. B. Trotter, David Chang, Fadi Towfic. Multi-omic Profiling of primary pancreatic adenocarcinomas obtained from the APACT adjuvant trial of nab-paclitaxel + gemcitabine vs gemcitabine [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2020 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2020;80(22 Suppl):Abstract nr PO-008.

Abstract A43: Spatial organization of pancreatic ductal adenocarcinoma (PDAC)–associated immune cells from the Adjuvant Pancreatic Adenocarcinoma Clinical Trial (APACT) study

Introduction: It is strikingly difficult to develop successful treatments for PDAC; even with curative resection, most patients die from early occult metastases. Prior studies identified the presence of tumor-infiltrating lymphocytes (TILs) in primary PDAC tumors as having prognostic significance in the PDAC adjuvant setting, sharpening the questions of what fraction of patients have immune-infiltrated tumors and what therapeutic strategies should be pursued in these patients vs. the non-infiltrated group. The phase 3 APACT trial evaluated the use of adjuvant nab-paclitaxel plus gemcitabine vs. gemcitabine in 866 patients with PDAC who had undergone primary tumor resection, with the primary endpoint of disease-free survival evaluated by independent review. We extended studies of the tumor microenvironment of PDAC to a large set of resected APACT primary tumors in an effort to further refine features of tumor or immune infiltrate that influence disease progression and to determine if chemotherapy regimen–specific predictive signatures are identifiable. Tissue analyses for a large subset of APACT samples included RNA-seq, DNA-seq, multiplexed immunohistochemistry (IHC), and proteomics.

Methods: We imaged and quantified markers for tumor cells, 7 different immune cells, and 2 immune checkpoint markers using bright-field chromogenic multiplexed IHC from pretreatment samples for more than 500 APACT primary tumor samples. We computationally defined the tumor, tumor margin, and distal stromal (&gt; 150 μm from tumor boundary) regions, and quantified densities and distributions of immune cells in these regions. As part of an initial analysis of more than 400 samples, we applied both unsupervised clustering and supervised classification to these IHC measurements to identify patient subgroups with similar spatial arrangements of immune cells relative to tumor regions.

Results: The preliminary analysis of normalized cell densities across all 3 tissue regions revealed 3 patient subgroups: one in which immune cells are mixed within the tumor regions; a second where immune cells approach the tumor boundary but are depleted within the tumor; and a third in which immune cells are depleted in both tumor and its margin, remaining at high densities only in the distal stromal regions. Within these latter subgroups, CD20+, CD4+, and CD8+ cells were more prevalently depleted from tumor and/or margin, whereas CD163+ and CD163+CMAF+ cells showed less of this arrangement. Nearly 85% of patients fell in the second or third patient group.

Conclusions: We are pursuing analyses of these data in conjunction with upcoming molecular and genetic profiling data to further elucidate the association of the immune cell populations and these subgroups with clinical outcomes. These data will provide an unprecedented opportunity for exploratory analysis and discovery of immune, molecular, and genetic biomarkers for PDAC patient stratification.

Citation Format: David J. Reiss, Thomas Lila, Suzana Couto, Sitharthan Kamalakaran, Yan Ren, Doug Bowman, Amber Ortiz, Maria Wang, Clifton Drew, Kao-Tai Tsai, Mathieu Marella, Brian Fox, Garth McGrath, Matthew Trotter, Fadi Towfic, Ian Cushman, Alexander Ratushny, Brian Lu, Daniel Pierce, Jim Cassidy. Spatial organization of pancreatic ductal adenocarcinoma (PDAC)–associated immune cells from the Adjuvant Pancreatic Adenocarcinoma Clinical Trial (APACT) study [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr A43.

Cereblon gene expression and correlation with clinical outcomes in patients with relapsed/refractory multiple myeloma treated with pomalidomide: an analysis of STRATUS

We analyzed gene expression levels of CRBN, cMYC, IRF4, BLIMP1, and XBP1 in 224 patients with multiple myeloma treated with pomalidomide and low-dose dexamethasone in the STRATUS study (ClinicalTrials.gov: NCT01712789; EudraCT number: 2012-001888-78). Clinical responses were observed at all CRBN expression levels. A trend in progression-free survival (PFS; p = .038) and a potential trend in overall survival (OS; p = .059) favoring high CRBN expressers were observed; however, no notable difference in overall response rate (ORR) was observed. ORR (30%), median PFS (17.7 weeks), and median OS (52.3 weeks) in low-CRBN expressers were comparable to those in the STRATUS intent-to-treat population (ORR, 33%; median PFS, 20.0 weeks; median OS, 51.7 weeks). A trend in ORR (p = .050) favoring higher cMYC expressers was observed with no notable difference in PFS or OS. This analysis does not support exploring CRBN as a biomarker for selecting patients for pomalidomide therapy.

A high-risk, Double-Hit, group of newly diagnosed myeloma identified by genomic analysis

Patients with newly diagnosed multiple myeloma (NDMM) with high-risk disease are in need of new treatment strategies to improve the outcomes. Multiple clinical, cytogenetic, or gene expression features have been used to identify high-risk patients, each of which has significant weaknesses. Inclusion of molecular features into risk stratification could resolve the current challenges. In a genome-wide analysis of the largest set of molecular and clinical data established to date from NDMM, as part of the Myeloma Genome Project, we have defined DNA drivers of aggressive clinical behavior. Whole-genome and exome data from 1273 NDMM patients identified genetic factors that contribute significantly to progression free survival (PFS) and overall survival (OS) (cumulative R² = 18.4% and 25.2%, respectively). Integrating DNA drivers and clinical data into a Cox model using 784 patients with ISS, age, PFS, OS, and genomic data, the model has a cumlative R² of 34.3% for PFS and 46.5% for OS. A high-risk subgroup was defined by recursive partitioning using either a) bi-allelic TP53 inactivation or b) amplification (≥4 copies) of CKS1B (1q21) on the background of International Staging System III, comprising 6.1% of the population (median PFS = 15.4 months; OS = 20.7 months) that was validated in an independent dataset. Double-Hit patients have a dire prognosis despite modern therapies and should be considered for novel therapeutic approaches.

Letter to the Editor response: Nygaard et al

The article by Nygaard and others (2016) proposes that applying batch correction approaches to microarray data from studies with unbalanced designs may inadvertently exaggerate the differences observed. In seeking to illustrate their point, Nygaard and others (2016) utilized a dataset (GSE61901) from a study we published (Towfic and others, 2014) and showed that one analysis pipeline utilizing the traditional approach to batch correction (ComBat) yielded over 1000 differentially expressed probesets, while an alternative approach proposed by Nygaard and others (2016). (utilizing batch as a fixed effect and averaging technical replicates) recovered 11 differentially expressed probesets.

Abstract 4725: Multiple Myeloma DREAM Challenge: A crowd-sourced challenge to improve identification of high-risk patients

Introduction: Multiple myeloma (MM) is a cancer of the plasma cells in the bone marrow, and its clinical course depends on a complex interplay of clinical traits and molecular characteristics of the plasma cells. Since risk-adapted therapy is becoming standard of care, there is an urgent need for a precise risk stratification model to assist in therapeutic decision-making. While progress has been made, there remains a significant opportunity to improve patient stratification to optimize treatment and to develop new therapies for high-risk patients. To accelerate the development and evaluation of such risk models in MM, we formed a DREAM Challenge, a crowd-sourced competition that engages large cross-disciplinary teams of experts to address complex problems in biomedicine.

Methods and Data: In collaboration with Multiple Myeloma Genome Project (MGP), clinical variables, patient outcomes, genetic, and gene expression data from thousands of samples were curated and harmonized from multiple public and private studies. In preparation for the challenge, a team of data scientists was assembled to evaluate this data, benchmark public high-risk models, and assess established prediction metrics with regard to progression-free survival (PFS) and overall survival (OS), the clinical endpoints of the challenge. Docker containers will be used to validate submitted models on private data that would otherwise not be available and to facilitate the transition of the best performing predictive signature to a clinical application. The MM DREAM challenge is accessible at: synapse.org.

Results: The international staging system (ISS) for myeloma was used as a baseline classifier for high-risk patients (PFS &lt; 18mo). We evaluated published high-risk signatures - UAMS-5, UAMS-17, UAMS-70, and EMC92 - as benchmarks and observed that they consistently outperformed the baseline ISS predictor. High-risk prediction scores from these models were moderately correlated, suggesting published classifiers capture non-overlapping determinants of risk. Development of de novo classifiers by our team integrating clinical and molecular data highlighted opportunities for model refinement and supports rationalization of a crowd-sourced challenge to advance the field.

Conclusion: Preliminary analysis of the challenge data suggests there is an opportunity to significantly improve risk stratification models in MM. In addition to the robust benchmarking of existing classifiers, we anticipate new, more accurate models will be proposed through a MM challenge given the scale of the combined data sets. We hope to uncover novel clinical and molecular traits that may yield insight into the pathology of MM and provide direction for follow-up studies. Importantly, this challenge will illustrate the advantages of leveraging public data and crowdsourcing to address therapeutically relevant questions in oncology. In addition, this challenge establishes a community resource for future research and benchmarking of novel classifiers.

Citation Format: Michael Mason, Michael Amatangelo, Daniel Auclair, Doug Bassett, Hongyue Dai, Andrew Dervan, Erin Flynt, Hartmut Goldschmidt, Dirk Hose, Konstantinos Mavrommatis, Gareth Morgan, Nikhil Munshi, Alex Ratushny, Dan Rozelle, Mehmet Samur, Frank Schmitz, Ken Shain, Anjan Thakurta, Fadi Towfic, Matthew Trotter, Brian Walker, Brian S. White, Thomas Yu, Justin Guinney. Multiple Myeloma DREAM Challenge: A crowd-sourced challenge to improve identification of high-risk patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4725. doi:10.1158/1538-7445.AM2017-4725

A pharmacogenetic signature of high response to Copaxone in late-phase clinical-trial cohorts of multiple sclerosis

Background

Copaxone is an efficacious and safe therapy that has demonstrated clinical benefit for over two decades in patients with relapsing forms of multiple sclerosis (MS). On an individual level, patients show variability in their response to Copaxone, with some achieving significantly higher response levels. The involvement of genes (e.g., HLA-DRB1*1501) with high inter-individual variability in Copaxone’s mechanism of action (MoA) suggests the potential contribution of genetics to treatment response. This study aimed to identify genetic variants associated with Copaxone response in patient cohorts from late-phase clinical trials.

Methods

Single nucleotide polymorphisms (SNPs) associated with high and low levels of response to Copaxone were identified using genome-wide SNP data in a discovery cohort of 580 patients from two phase III clinical trials of Copaxone. Multivariable Bayesian modeling on the resulting SNPs in an expanded discovery cohort with 1171 patients identified a multi-SNP signature of Copaxone response. This signature was examined in 941 Copaxone-treated MS patients from seven independent late-phase trials of Copaxone and assessed for specificity to Copaxone in 310 Avonex-treated and 311 placebo-treated patients, also from late-phase trials.

Results

A four-SNP signature consisting of rs80191572 (in UVRAG), rs28724893 (in HLA-DQB2), rs1789084 (in MBP), and rs139890339 (in ZAK(CDCA7)) was identified as significantly associated with Copaxone response. Copaxone-treated signature-positive patients had a greater reduction in annualized relapse rate (ARR) compared to signature-negative patients in both discovery and independent cohorts, an effect not observed in Avonex-treated patients. Additionally, signature-positive placebo-treated cohorts did not show a reduction in ARR, demonstrating the predictive as opposed to prognostic nature of the signature. A 10% subset of patients, delineated by the signature, showed marked improvements across multiple clinical parameters, including ARR, MRI measures, and higher proportion with no evidence of disease activity (NEDA).

Conclusions

This study is the largest pharmacogenetic study in MS reported to date. Gene regions underlying the four-SNP signature have been linked with pathways associated with either Copaxone’s MoA or the pathophysiology of MS. The pronounced association of the four-SNP signature with clinical improvements in a ~10% subset of the MS patient population demonstrates the complex interplay of immune mechanisms and the individualized nature of response to Copaxone.

Electronic supplementary material

The online version of this article (doi:10.1186/s13073-017-0436-y) contains supplementary material, which is available to authorized users.

Pridopidine activates neuroprotective pathways impaired in Huntington Disease

Pridopidine has demonstrated improvement in Huntington Disease (HD) motor symptoms as measured by secondary endpoints in clinical trials. Originally described as a dopamine stabilizer, this mechanism is insufficient to explain the clinical and preclinical effects of pridopidine. This study therefore explored pridopidine’s potential mechanisms of action. The effect of pridopidine versus sham treatment on genome-wide expression profiling in the rat striatum was analysed and compared to the pathological expression profile in Q175 knock-in (Q175 KI) vs Q25 WT mouse models. A broad, unbiased pathway analysis was conducted, followed by testing the enrichment of relevant pathways. Pridopidine upregulated the BDNF pathway (P = 1.73E-10), and its effect on BDNF secretion was sigma 1 receptor (S1R) dependent. Many of the same genes were independently found to be downregulated in Q175 KI mice compared to WT (5.2e-7 < P < 0.04). In addition, pridopidine treatment upregulated the glucocorticoid receptor (GR) response, D1R-associated genes and the AKT/PI3K pathway (P = 1E-10, P = 0.001, P = 0.004, respectively). Pridopidine upregulates expression of BDNF, D1R, GR and AKT/PI3K pathways, known to promote neuronal plasticity and survival, as well as reported to demonstrate therapeutic benefit in HD animal models. Activation of S1R, necessary for its effect on the BDNF pathway, represents a core component of the mode of action of pridopidine. Since the newly identified pathways are downregulated in neurodegenerative diseases, including HD, these findings suggest that pridopidine may exert neuroprotective effects beyond its role in alleviating some symptoms of HD.

Abstract 789: Leveraging transcriptomic and genomic data to better select models for preclinical oncology therapeutic development to identify cell lines most similar to patient tumors

Cancer cell lines represent the front line of new compound testing, and results from these experiments often decide which compounds go on for further testing. Genomic context plays a critical role in drug response and now genomic data for tumors and cell lines are widely available. However, cell lines are often chosen based on ease of access, literature prevalence, and ease of culture. We combined gene expression and CNV/mutation profiling from four pancreatic cancer tumor datasets (GSE21501, GSE28735, ICGC, TCGA,) and three pancreatic cancer cell line datasets (Klijn et al, Collisson et al, and CCLE) to identify which cell lines best match patient tumors.

CNV comparison revealed that popular cell lines do not always have the best CNV correlation with tumors: when comparing pancreatic cancer tumors to cell lines, the citations of the top five cell lines by CNV correlation were less than 10% of the pancreatic cancer cell line total. Next we filtered for driver mutations including SMAD4 and CDKN2A using mutation scoring algorithms and clustered tumors and cell lines. We found that many cell lines with few citation counts clustered readily amongst tumors (such as L33). Leveraging the hypothesis that different hits in the same pathway can have a similar downstream effect, we combined CNV, expression and mutation data and clustered cell lines together with tumors based on overall aberrations in MSigDB cancer pathways. L33 and YAPC clustered near tumors while the majority of other cell lines clustered together.

To identify coexpressed gene clusters, we ran WGCNA individually in all seven datasets and discovered modules consistent in cell line and tumor datasets using iGraph. One of the most interesting modules (interferon regulated genes) is expressed highly in the majority of tumors profiled. About half of cell lines also express this module highly, suggesting that they may be more ideal models for high interferon expression tumors than other cell lines.

Here we present evidence demonstrating that certain cell lines mimic pancreatic tumor genomes more closely while others represent patterns of genomic features not commonly observed in vivo. We also show that certain biologically relevant tumor subtypes may be better represented by some cell lines than others. Our analysis highlights the emerging role of genomics in advancing the clinical success of therapeutic trials.

Citation Format: Yoonjeong Cha, Adam Labradorf, Joseph Perez-Rogers, Brian Haas, Andrew Lysaght, Brian Weiner, Fadi Towfic, Kevin Fowler, Benjamin Zeskind, Sarah Kolitz, Badri Vardarajan, Maxim Artyomov, Rebecca L. Kusko. Leveraging transcriptomic and genomic data to better select models for preclinical oncology therapeutic development to identify cell lines most similar to patient tumors. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 789.

Abstract A78: Leveraging genomics to optimize models for accelerated pancreatic cancer drug development

Cell lines used for pre-clinical testing of oncology compounds are not always chosen based on how well they models patient tumors. Instead they are often chosen based on availability and literature prevalence. The advent of high throughput genomic profiling demonstrates a causative relationship between genomic features and drug response, suggesting that cancer drug discovery could be accelerated by using genomics as a criteria to find ideal cell lines for a given cancer type. The overall oncology clinical trial success rate is dismally low, especially in pancreatic cancer. Pancreatic cancer has a five year survival of 5-6% and is predicted to be the second leading cause of cancer by 2030 with a dearth of promising medicines currently in trials. In order to forecast optimal cell lines for drug testing in pancreatic cancer, we leveraged gene expression, mutations, and copy number variation (CNV) data to compare tumors from The Cancer Genome Atlas (TCGA) to cell lines in Cancer Cell Line Encyclopedia (CCLE). To approximate cell line usage, the number of hits for each cell line in PubMed and Google Scholar were combined. Less than 20% of queried pancreatic cancer cell lines represented more than 88% of the total search hits, demonstrating a robust bias towards certain cell lines. We calculated the CNV correlation between each cell line and each tumor. The cell lines that were popular in literature, such as DAN-G (24% of citations), were often ranked worst by CNV correlation with tumors while some cell lines which were rarely cited such as L33 had among the highest CNV correlation. Next, we filtered mutation data using publicly available mutation scoring algorithms to select the most cancer driving mutations. Hierarchical clustering was applied to the tumor samples and cell lines together based on the presence or absence of the top scoring mutations in order to pinpoint cell lines with mutational spectra similar to tumors. In support of observations made in CNV data, popular cell lines such as DAN-G clustered with other cell lines and L33 clustered predominantly amongst tumor samples, providing further evidence that L33 may be an ideal cell line for modeling pancreatic cancer drug response. In order to leverage all available data types, the selected CNVs and mutations were combined into a pathway level event matrix based on the number of relevant mutations or CNVs within a given pathway and then clustered. Unsurprisingly these results show that most cell lines are much more similar to each other than to tumors. However, a few cell lines (including L33) cluster with tumor samples. Overall our results demonstrate that comprehensively L33 shows the best similarity to pancreatic cancer tumors. We believe that selecting preclinical screening methods that best match relevant tumor biology and genomic drivers could help accelerate the development of new medicines for a variety of cancers.

Citation Format: Yoonjeong Cha, Andrew Lysaght, Brian Weiner, Sarah Kolitz, Fadi Towfic, Kevin Fowler, Badri Vardarajan, Maxim Artyomov, Benjamin Zeskind, Rebecca Kusko. Leveraging genomics to optimize models for accelerated pancreatic cancer drug development. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A78.

Leveraging existing data sets to generate new insights into Alzheimer's disease biology in specific patient subsets

To generate new insights into the biology of Alzheimer’s Disease (AD), we developed methods to combine and reuse a wide variety of existing data sets in new ways. We first identified genes consistently associated with AD in each of four separate expression studies, and confirmed this result using a fifth study. We next developed algorithms to search hundreds of thousands of Gene Expression Omnibus (GEO) data sets, identifying a link between an AD-associated gene (NEUROD6) and gender. We therefore stratified patients by gender along with APOE4 status, and analyzed multiple SNP data sets to identify variants associated with AD. SNPs in either the region of NEUROD6 or SNAP25 were significantly associated with AD, in APOE4+ females and APOE4+ males, respectively. We developed algorithms to search Connectivity Map (CMAP) data for medicines that modulate AD-associated genes, identifying hypotheses that warrant further investigation for treating specific AD patient subsets. In contrast to other methods, this approach focused on integrating multiple gene expression datasets across platforms in order to achieve a robust intersection of disease-affected genes, and then leveraging these results in combination with genetic studies in order to prioritize potential genes for targeted therapy.

Abstract 4741: Improving pancreatic cancer drug discovery by leveraging genomics to select better in vitro models

Currently, pancreatic cancer has an estimated 5-year survival rate of only 5-6%. The projection that pancreatic cancer will be the second leading cause of cancer related death by 2020 compounded by the numerous clinical trial failures precipitates the need for novel approaches to accelerate progress in new medicine development. Cell lines used for screening pre-clinical compounds prior to animal models and human testing are usually chosen based on ease of access and literature prevalence. However, the constellation of genomic derangements in cell lines commonly used for in vitro studies may not be representative of pancreatic cancer. In this study, we leveraged copy number variation (CNV) and targeted sequencing data from The Cancer Genome Atlas (TCGA) and the Cancer Cell Line Encyclopedia (CCLE) to predict optimal cell lines that mirror pancreatic cancer genomes most closely. We calculated the frequency of each CCLE pancreatic cancer cell line in literature and compared this to how well each cell line recapitulates the pancreatic cancer population. Unsurprisingly, we observed that CCLE pancreatic cancer cell lines overall have more frequent CNVs and mutations than TCGA pancreatic cancer tumors. This observation is likely due to inherent genomic instability of cell lines and underscores the importance of using low passage cells. Next, we directly compared the median per gene CNV values in TCGA pancreatic cancer tumors and pancreatic cancer cell lines in CCLE. Contrary to our expectation, the top five cell lines by CNV correlation with TCGA pancreatic tumors represented only 6% out of all literature search hits for all CCLE pancreatic cancer cell lines, indicating the availability of more optimal cell lines from a genomics perspective. Additionally, we leveraged targeted sequencing data to compare the most frequent mutations with medium to high Mutation Assessor scores in TCGA pancreatic cancer tumors to CCLE pancreatic cancer cell lines. The seven most common mutations by this method in TCGA pancreatic cancer tumors were: KRAS, TP53, MYH8, TAOK2, PCDH15, ATRX, and CDKN2A. Using hierarchical clustering based on the presence or absence of these 7 mutations in pancreatic cancer CCLE cell lines and TCGA tumors, we showed that some cell lines readily clustered amongst TCGA tumors (such as BXPC3), while others occupied discrete branches of the dendrogram exclusive of most TCGA tumors such as PK1 and PANC1. This implies that while some cell line mimic pancreatic tumor mutations closely, others represent mutation constellations not commonly observed in patients. It is possible to apply this method to other cancer types, given consideration for potentially different cancer biology. In summary, our work reports that many popular pancreatic cancer cell lines harbor distinct genomic aberration profiles from pancreatic cancer tumors and highlights the emerging role of genomics in advancing the clinical success of therapeutic trials.

Citation Format: Yoonjeong Cha, Andrew Lysaght, Rain Cui, Brian Weiner, Sarah Kolitz, Fadi Towfic, Jason Funt, Kevin Fowler, Badri Vardarajan, Maxim Artyomov, Benjamin Zeskind, Rebecca Kusko. Improving pancreatic cancer drug discovery by leveraging genomics to select better in vitro models. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4741. doi:10.1158/1538-7445.AM2015-4741

Gene expression studies of a human monocyte cell line identify dissimilarities between differently manufactured glatiramoids

Glatiramer Acetate (GA) has provided safe and effective treatment for multiple sclerosis (MS) patients for two decades. It acts as an antigen, yet the precise mechanism of action remains to be fully elucidated, and no validated pharmacokinetic or pharmacodynamic biomarkers exist. In order to better characterize GA’s biological impact, genome-wide expression studies were conducted with a human monocyte (THP-1) cell line. Consistent with previous literature, branded GA upregulated anti-inflammatory markers (e.g. IL10), and modulated multiple immune-related pathways. Despite some similarities, significant differences were observed between expression profiles induced by branded GA and Probioglat, a differently-manufactured glatiramoid purported to be a generic GA. Key results were verified using qRT-PCR. Genes (e.g. CCL5, adj. p < 4.1 × 10⁻⁵) critically involved in pro-inflammatory pathways (e.g. response to lipopolysaccharide, adj. p = 8.7 × 10⁻⁴) were significantly induced by Probioglat compared with branded GA. Key genes were also tested and confirmed at the protein level, and in primary human monocytes. These observations suggest differential biological impact by the two glatiramoids and warrant further investigation.

Use of genetic technologies to compare medicines

In order to ensure that patients receive the safest and most effective medicines possible, it is often necessary to compare medicines and assess the extent to which they are similar in their clinical impact. Full clinical trials with appropriate endpoints remain the only method to compare the clinical impact of two medicines with absolute certainty. Other available methods (including physicochemical analysis, genomics, and transcriptomics) can provide partial information about certain aspects of a medicine's biological impact, with possible clinical implications. Especially for biologics and non-biological complex drugs, which are more difficult to characterize by physicochemical means than small molecules, genomics and transciptomic studies can yield valuable insights for physicians, regulators, and drug developers. In this review, we cite and summarize a variety of studies that exemplify the emerging science of applying genomics and transcriptomics technologies to compare medicines. We discuss key aspects of experimental design, conduct of genetic assays, and advanced data analysis, all of which are critical for the successful execution of such studies. Finally, we propose new areas for which such studies can be applied to maximize patient benefit and reduce safety issues.

Comparing the biological impact of glatiramer acetate with the biological impact of a generic

For decades, policies regarding generic medicines have sought to provide patients with economical access to safe and effective drugs, while encouraging the development of new therapies. This balance is becoming more challenging for physicians and regulators as biologics and non-biological complex drugs (NBCDs) such as glatiramer acetate demonstrate remarkable efficacy, because generics for these medicines are more difficult to assess. We sought to develop computational methods that use transcriptional profiles to compare branded medicines to generics, robustly characterizing differences in biological impact. We combined multiple computational methods to determine whether differentially expressed genes result from random variation, or point to consistent differences in biological impact of the generic compared to the branded medicine. We applied these methods to analyze gene expression data from mouse splenocytes exposed to either branded glatiramer acetate or a generic. The computational methods identified extensive evidence that branded glatiramer acetate has a more consistent biological impact across batches than the generic, and has a distinct impact on regulatory T cells and myeloid lineage cells. In summary, we developed a computational pipeline that integrates multiple methods to compare two medicines in an innovative way. This pipeline, and the specific findings distinguishing branded glatiramer acetate from a generic, can help physicians and regulators take appropriate steps to ensure safety and efficacy.

Genetic variants synthesize to produce paneth cell phenotypes that define subtypes of Crohn's disease

BACKGROUND & AIMS

Genetic susceptibility loci for Crohn's disease (CD) are numerous, complex, and likely interact with undefined components of the environment. It has been a challenge to link the effects of particular loci to phenotypes of cells associated with pathogenesis of CD, such as Paneth cells. We investigated whether specific phenotypes of Paneth cells associated with particular genetic susceptibility loci can be used to define specific subtypes of CD.

METHODS

We performed a retrospective analysis of 119 resection specimens collected from patients with CD at 2 separate medical centers. Paneth cell phenotypes were classified as normal or abnormal (with disordered, diminished, diffuse, or excluded granule phenotypes) based on lysozyme-positive secretory granule morphology. To uncover the molecular basis of the Paneth cell phenotypes, we developed methods to determine transcriptional profiles from whole-thickness and laser-capture microdissected, formalin-fixed, paraffin-embedded tissue sections.

RESULTS

The proportion of abnormal Paneth cells was associated with the number of CD-associated NOD2 risk alleles. The cumulative number of NOD2 and ATG16L1 risk alleles had an additive effect on the proportion of abnormal Paneth cells. Unsupervised clustering analysis of demographic and Paneth cell data divided patients into 2 principal subgroups, defined by high and low proportions of abnormal Paneth cells. The disordered and diffuse abnormal Paneth cell phenotypes were associated with an altered transcriptional signature of immune system activation. We observed an inverse correlation between abnormal Paneth cells and presence of granuloma. In addition, high proportions of abnormal Paneth cells were associated with shorter time to disease recurrence after surgery.

CONCLUSIONS

Histologic analysis of Paneth cell phenotypes can be used to divide patients with CD into subgroups with distinct pathognomonic and clinical features.

Defining the transcriptional and cellular landscape of type 1 diabetes in the NOD mouse

Our ability to successfully intervene in disease processes is dependent on definitive diagnosis. In the case of autoimmune disease, this is particularly challenging because progression of disease is lengthy and multifactorial. Here we show the first chronological compendium of transcriptional and cellular signatures of diabetes in the non-obese diabetic mouse. Our data relates the immunological environment of the islets of Langerhans with the transcriptional profile at discrete times. Based on these data, we have parsed diabetes into several discrete phases. First, there is a type I interferon signature that precedes T cell activation. Second, there is synchronous infiltration of all immunological cellular subsets and a period of control. Finally, there is the killing phase of the diabetogenic process that is correlated with an NF-kB signature. Our data provides a framework for future examination of autoimmune diabetes and its disease progression markers.

B-cell ligand processing pathways detected by large-scale comparative analysis

The initiation of B-cell ligand recognition is a critical step for the generation of an immune response against foreign bodies. We sought to identify the biochemical pathways involved in the B-cell ligand recognition cascade and sets of ligands that trigger similar immunological responses. We utilized several comparative approaches to analyze the gene coexpression networks generated from a set of microarray experiments spanning 33 different ligands. First, we compared the degree distributions of the generated networks. Second, we utilized a pairwise network alignment algorithm, BiNA, to align the networks based on the hubs in the networks. Third, we aligned the networks based on a set of KEGG pathways. We summarized our results by constructing a consensus hierarchy of pathways that are involved in B cell ligand recognition. The resulting pathways were further validated through literature for their common physiological responses. Collectively, the results based on our comparative analyses of degree distributions, alignment of hubs, and alignment based on KEGG pathways provide a basis for molecular characterization of the immune response states of B-cells and demonstrate the power of comparative approaches (e.g., gene coexpression network alignment algorithms) in elucidating biochemical pathways involved in complex signaling events in cells.

Protein-RNA interface residue prediction using machine learning: an assessment of the state of the art

Background

RNA molecules play diverse functional and structural roles in cells. They function as messengers for transferring genetic information from DNA to proteins, as the primary genetic material in many viruses, as catalysts (ribozymes) important for protein synthesis and RNA processing, and as essential and ubiquitous regulators of gene expression in living organisms. Many of these functions depend on precisely orchestrated interactions between RNA molecules and specific proteins in cells. Understanding the molecular mechanisms by which proteins recognize and bind RNA is essential for comprehending the functional implications of these interactions, but the recognition ‘code’ that mediates interactions between proteins and RNA is not yet understood. Success in deciphering this code would dramatically impact the development of new therapeutic strategies for intervening in devastating diseases such as AIDS and cancer. Because of the high cost of experimental determination of protein-RNA interfaces, there is an increasing reliance on statistical machine learning methods for training predictors of RNA-binding residues in proteins. However, because of differences in the choice of datasets, performance measures, and data representations used, it has been difficult to obtain an accurate assessment of the current state of the art in protein-RNA interface prediction.

Results

We provide a review of published approaches for predicting RNA-binding residues in proteins and a systematic comparison and critical assessment of protein-RNA interface residue predictors trained using these approaches on three carefully curated non-redundant datasets. We directly compare two widely used machine learning algorithms (Naïve Bayes (NB) and Support Vector Machine (SVM)) using three different data representations in which features are encoded using either sequence- or structure-based windows. Our results show that (i) Sequence-based classifiers that use a position-specific scoring matrix (PSSM)-based representation (PSSMSeq) outperform those that use an amino acid identity based representation (IDSeq) or a smoothed PSSM (SmoPSSMSeq); (ii) Structure-based classifiers that use smoothed PSSM representation (SmoPSSMStr) outperform those that use PSSM (PSSMStr) as well as sequence identity based representation (IDStr). PSSMSeq classifiers, when tested on an independent test set of 44 proteins, achieve performance that is comparable to that of three state-of-the-art structure-based predictors (including those that exploit geometric features) in terms of Matthews Correlation Coefficient (MCC), although the structure-based methods achieve substantially higher Specificity (albeit at the expense of Sensitivity) compared to sequence-based methods. We also find that the expected performance of the classifiers on a residue level can be markedly different from that on a protein level. Our experiments show that the classifiers trained on three different non-redundant protein-RNA interface datasets achieve comparable cross-validation performance. However, we find that the results are significantly affected by differences in the distance threshold used to define interface residues.

Conclusions

Our results demonstrate that protein-RNA interface residue predictors that use a PSSM-based encoding of sequence windows outperform classifiers that use other encodings of sequence windows. While structure-based methods that exploit geometric features can yield significant increases in the Specificity of protein-RNA interface residue predictions, such increases are offset by decreases in Sensitivity. These results underscore the importance of comparing alternative methods using rigorous statistical procedures, multiple performance measures, and datasets that are constructed based on several alternative definitions of interface residues and redundancy cutoffs as well as including evaluations on independent test sets into the comparisons.

BioNetwork Bench: Database and Software for Storage, Query, and Analysis of Gene and Protein Networks

Gene and protein networks offer a powerful approach for integration of the disparate yet complimentary types of data that result from high-throughput analyses. Although many tools and databases are currently available for accessing such data, they are left unutilized by bench scientists as they generally lack features for effective analysis and integration of both public and private datasets and do not offer an intuitive interface for use by scientists with limited computational expertise. We describe BioNetwork Bench, an open source, user-friendly suite of database and software tools for constructing, querying, and analyzing gene and protein network models. It enables biologists to analyze public as well as private gene expression; interactively query gene expression datasets; integrate data from multiple networks; store and selectively share the data and results. Finally, we describe an application of BioNetwork Bench to the assembly and iterative expansion of a gene network that controls the differentiation of retinal progenitor cells into rod photoreceptors. The tool is available from http://bionetworkbench.sourceforge.net/

Exploration of the relationship between topology and designability of conformations

Protein structures are evolutionarily more conserved than sequences, and sequences with very low sequence identity frequently share the same fold. This leads to the concept of protein designability. Some folds are more designable and lots of sequences can assume that fold. Elucidating the relationship between protein sequence and the three-dimensional (3D) structure that the sequence folds into is an important problem in computational structural biology. Lattice models have been utilized in numerous studies to model protein folds and predict the designability of certain folds. In this study, all possible compact conformations within a set of two-dimensional and 3D lattice spaces are explored. Complementary interaction graphs are then generated for each conformation and are described using a set of graph features. The full HP sequence space for each lattice model is generated and contact energies are calculated by threading each sequence onto all the possible conformations. Unique conformation giving minimum energy is identified for each sequence and the number of sequences folding to each conformation (designability) is obtained. Machine learning algorithms are used to predict the designability of each conformation. We find that the highly designable structures can be distinguished from other non-designable conformations based on certain graphical geometric features of the interactions. This finding confirms the fact that the topology of a conformation is an important determinant of the extent of its designability and suggests that the interactions themselves are important for determining the designability.

Detection of gene orthology from gene co-expression and protein interaction networks

BACKGROUND

Ortholog detection methods present a powerful approach for finding genes that participate in similar biological processes across different organisms, extending our understanding of interactions between genes across different pathways, and understanding the evolution of gene families.

RESULTS

We exploit features derived from the alignment of protein-protein interaction networks and gene-coexpression networks to reconstruct KEGG orthologs for Drosophila melanogaster, Saccharomyces cerevisiae, Mus musculus and Homo sapiens protein-protein interaction networks extracted from the DIP repository and Mus musculus and Homo sapiens and Sus scrofa gene coexpression networks extracted from NCBI's Gene Expression Omnibus using the decision tree, Naive-Bayes and Support Vector Machine classification algorithms.

CONCLUSIONS

The performance of our classifiers in reconstructing KEGG orthologs is compared against a basic reciprocal BLAST hit approach. We provide implementations of the resulting algorithms as part of BiNA, an open source biomolecular network alignment toolkit.