Techniques to produce and evaluate realistic multivariate synthetic data

Data modeling requires a sufficient sample size for reproducibility. A small sample size can inhibit model evaluation. A synthetic data generation technique addressing this small sample size problem is evaluated: from the space of arbitrarily distributed samples, a subgroup (class) has a latent multivariate normal characteristic; synthetic data can be generated from this class with univariate kernel density estimation (KDE); and synthetic samples are statistically like their respective samples. Three samples (n = 667) were investigated with 10 input variables (X). KDE was used to augment the sample size in X. Maps produced univariate normal variables in Y. Principal component analysis in Y produced uncorrelated variables in T, where the probability density functions were approximated as normal and characterized; synthetic data was generated with normally distributed univariate random variables in T. Reversing each step produced synthetic data in Y and X. All samples were approximately multivariate normal in Y, permitting the generation of synthetic data. Probability density function and covariance comparisons showed similarity between samples and synthetic samples. A class of samples has a latent normal characteristic. For such samples, this approach offers a solution to the small sample size problem. Further studies are required to understand this latent class.

PATH-SURVEYOR: pathway level survival enquiry for immuno-oncology and drug repurposing

Pathway-level survival analysis offers the opportunity to examine molecular pathways and immune signatures that influence patient outcomes. However, available survival analysis algorithms are limited in pathway-level function and lack a streamlined analytical process. Here we present a comprehensive pathway-level survival analysis suite, PATH-SURVEYOR, which includes a Shiny user interface with extensive features for systematic exploration of pathways and covariates in a Cox proportional-hazard model. Moreover, our framework offers an integrative strategy for performing Hazard Ratio ranked Gene Set Enrichment Analysis and pathway clustering. As an example, we applied our tool in a combined cohort of melanoma patients treated with checkpoint inhibition (ICI) and identified several immune populations and biomarkers predictive of ICI efficacy. We also analyzed gene expression data of pediatric acute myeloid leukemia (AML) and performed an inverse association of drug targets with the patient's clinical endpoint. Our analysis derived several drug targets in high-risk KMT2A-fusion-positive patients, which were then validated in AML cell lines in the Genomics of Drug Sensitivity database. Altogether, the tool offers a comprehensive suite for pathway-level survival analysis and a user interface for exploring drug targets, molecular features, and immune populations at different resolutions.

Abstract 6029: Multi-omic landscape of squamous cell lung cancer

Patients with squamous cell lung cancer (SCC) have high unmet medical need. Knowledge of these tumors is limited, and a lack of targetable genomic drivers means patients have few treatment options. To provide a detailed analysis on the influence of genomic alterations to proteome-level changes in SCC, we previously integrated DNA copy number, somatic mutations, RNA-sequencing, and expression proteomics in a cohort of 108 SCC patients. A major finding was identification of three proteomic subtypes, two of which made up the majority (87%) of tumors: the “Inflamed” subtype was enriched for B-cell rich tertiary lymphoid structures (TLS), and the “Redox” subtype was enriched for redox pathways and NFE2L2/KEAP1 alterations but had significantly less immune infiltration. We hypothesized these proteomic subtypes would give rise to distinct metabolic signatures. Therefore, we performed untargeted metabolomics on 87 tumors from the same cohort using chromatographic separation on a HILIC column, followed by analysis on a Q Exactive HF mass spectrometer. This analysis yielded 7,344 features corresponding to 7,072 unannotated metabolites and 272 identified metabolites. Glutathione, a key redox metabolite, was anticorrelated with immune score (R = -0.44, padj = 0.004) calculated from our transcriptomic data with the ESTIMATE algorithm, and glutathione was elevated in the Redox proteomic subtype (0.58 log2 ratio, padj = 9.87E-04). Consensus clustering was next used to identify novel metabolomic subtypes of SCC. Surprisingly, none of the five metabolomic subtypes we identified corresponded to proteomic subtype or NFE2L2/KEAP1 alteration (Fisher’s Exact test p-values > 0.05). The fifth subtype had 332 metabolites (26 identified) differentially expressed (> 1.5 fold-change, padj < 0.05) with ascorbate and aldarate metabolism as the top enriched pathway (padj = 3.36E-04). Interestingly, this fifth metabolomic subtype had significantly higher DNp63-alpha (p = 2.40E-05), a primary transcript of delta-N p63 that is known to promote non-small cell lung cancer. Ongoing integrative analyses across omic types will determine how p53, p63, and p73 transcripts influence these metabolomic subtypes, how these transcripts relate to the poor immune infiltration in some SCC tumors, and if these transcripts relate to novel metabolic vulnerabilities in SCC.

Citation Format: Paul Stewart, Ashley Lui, Eric Welsh, Dalia Ercan, Vanessa Rubio, Hayley Ackerman, Guohui Li, Bin Fang, Steven Eschrich, John Koomen, Elsa Flores, Eric Haura, Gina DeNicola. Multi-omic landscape of squamous cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6029.

DRPPM-PATH-SURVEIOR: Plug-and-Play Survival Analysis of Pathway-level Signatures and Immune Components

Pathway-level survival analysis offers the opportunity to examine molecular pathways and immune signatures that influence patient outcomes. However, available survival analysis algorithms are limited in pathway-level function and lack a streamlined analytical process. Here we present a comprehensive pathway-level survival analysis suite, DRPPM-PATH-SURVEIOR, which includes a Shiny user interface with extensive features for systematic exploration of pathways and covariates in a Cox proportional-hazard model. Moreover, our framework offers an integrative strategy for performing Hazard Ratio ranked Gene Set Enrichment Analysis (GSEA) and pathway clustering. As an example, we applied our tool in a combined cohort of melanoma patients treated with checkpoint inhibition (ICI) and identified several immune populations and biomarkers predictive of ICI efficacy. We also analyzed gene expression data of pediatric acute myeloid leukemia (AML) and performed an inverse association of drug targets with the patient's clinical endpoint. Our analysis derived several drug targets in high-risk KMT2A-fusion-positive patients, which were then validated in AML cell lines in the Genomics of Drug Sensitivity database. Altogether, the tool offers a comprehensive suite for pathway-level survival analysis and a user interface for exploring drug targets, molecular features, and immune populations at different resolutions.

Ethnic and racial-specific differences in levels of centrosome-associated mitotic kinases, proliferative and epithelial-to-mesenchymal markers in breast cancers

Molecular epidemiology evidence indicates racial and ethnic differences in the aggressiveness and survival of breast cancer. Hispanics/Latinas (H/Ls) and non-Hispanic Black women (NHB) are at higher risk of breast cancer (BC)-related death relative to non-Hispanic white (NHW) women in part because they are diagnosed with hormone receptor-negative (HR) subtype and at higher stages. Since the cell cycle is one of the most commonly deregulated cellular processes in cancer, we propose that the mitotic kinases TTK (or Mps1), TBK1, and Nek2 could be novel targets to prevent breast cancer progression among NHBs and H/Ls. In this study, we calculated levels of TTK, p-TBK1, epithelial (E-cadherin), mesenchymal (Vimentin), and proliferation (Ki67) markers through immunohistochemical (IHC) staining of breast cancer tissue microarrays (TMAs) that includes samples from 6 regions in the Southeast of the United States and Puerto Rico -regions enriched with NHB and H/L breast cancer patients. IHC analysis showed that TTK, Ki67, and Vimentin were significantly expressed in triple-negative (TNBC) tumors relative to other subtypes, while E-cadherin showed decreased expression. TTK correlated with all of the clinical variables but p-TBK1 did not correlate with any of them. TCGA analysis revealed that the mRNA levels of multiple mitotic kinases, including TTK, Nek2, Plk1, Bub1, and Aurora kinases A and B, and transcription factors that are known to control the expression of these kinases (e.g. FoxM1 and E2F1-3) were upregulated in NHBs versus NHWs and correlated with higher aneuploidy indexes in NHB, suggesting that these mitotic kinases may be future novel targets for breast cancer treatment in NHB women.

Genome-wide host methylation profiling of anal and cervical carcinoma

HPV infection results in changes in host gene methylation which, in turn, are thought to contribute to the neoplastic progression of HPV-associated cancers. The objective of this study was to identify joint and disease-specific genome-wide methylation changes in anal and cervical cancer as well as changes in high-grade pre-neoplastic lesions. Formalin-fixed paraffin-embedded (FFPE) anal tissues (n = 143; 99% HPV+) and fresh frozen cervical tissues (n = 28; 100% HPV+) underwent microdissection, DNA extraction, HPV genotyping, bisulfite modification, DNA restoration (FFPE) and analysis by the Illumina HumanMethylation450 Array. Differentially methylated regions (DMR; t test q<0.01, 3 consecutive significant CpG probes and mean Δβ methylation value>0.3) were compared between normal and cancer specimens in partial least squares (PLS) models and then used to classify anal or cervical intraepithelial neoplasia-3 (AIN3/CIN3). In AC, an 84-gene PLS signature (355 significant probes) differentiated normal anal mucosa (NM; n = 9) from AC (n = 121) while a 36-gene PLS signature (173 significant probes) differentiated normal cervical epithelium (n = 10) from CC (n = 9). The CC progression signature was validated using three independent publicly available datasets (n = 424 cases). The AC and CC progression PLS signatures were interchangeable in segregating normal, AIN3/CIN3 and AC and CC and were found to include 17 common overlapping hypermethylated genes. Moreover, these signatures segregated AIN3/CIN3 lesions similarly into cancer-like and normal-like categories. Distinct methylation changes occur across the genome during the progression of AC and CC with overall similar profiles and add to the evidence suggesting that HPV-driven oncogenesis may result in similar non-random methylomic events. Our findings may lead to identification of potential epigenetic drivers of HPV-associated cancers and also, of potential markers to identify higher risk pre-cancerous lesions.

Effects of long-term norepinephrine treatment on normal immortalized ovarian and fallopian tube cells

Sustained adrenergic stimulation by norepinephrine (NE) contributes to ovarian carcinoma metastasis and impairment of chemotherapy response. Although the effect of sustained NE stimulation in cancer progression is well established, less is known about its role in cancer initiation. To determine the extent to which stress hormones influence ovarian cancer initiation, we conducted a long-term (> 3 months; > 40 population doublings) experiment in which normal immortalized fallopian tube secretory (iFTSEC283) and ovarian surface epithelial (iOSE11) cell lines and their isogenic pairs containing a p53 mutation (iFTSEC283p53R175H; iOSE11p53R175H), were continuously exposed to NE (100 nM, 1 μM, 10 μM). Fallopian tube cells displayed a p53-independent increase in proliferation and colony-forming ability in response to NE, while ovarian surface epithelial cells displayed a p53-independent decrease in both assays. Fallopian tube cells with mutant p53 showed a mild loss of chromosomes and TP53 status was also a defining factor in transcriptional response of fallopian tube cells to long-term NE treatment.

TAMI-54. THE PRESENCE OF IMMUNE CELL INFILTRATES IN THE TISSUE MICROENVIRONMENT OF HIGH-GRADE GLIOMAS AND THEIR ASSOCIATION WITH OVERALL SURVIVAL

OBJECTIVE

Tumor-associated microglia and macrophages (TAMs) influence brain tumor biology and promote tumor-proliferating phenotype. Studies have shown these cell types predict outcomes in various tumor sites with limited evidence in high-grade gliomas (HGG). In this study, we assessed the presence of immune cell infiltrate (ICI) and overall survival (OS) in HGGs.

METHODS

A total of 97 consecutively treated patients with primary HGG with complete gene expression profiling were identified from our IRB-approved institutional tissue biorepository. Patients underwent primary surgical resection between 02/2004 and 03/2011. Gene expression levels were assessed by Affymetrix Hu-RSTA assays (Affymetrix; Santa Clara, CA). CIBERSORT estimated the presence of ICI via gene expression deconvolution. Tumor characteristics and the presence of 22 individual ICI subtypes were assessed with respect to OS. Time-to-event analyses were performed with Kaplan-Meier estimates and compared via log-rank test. Associations between the ICI and outcomes were explored using Cox-regression. P&lt; 0.05 (two-tailed) was considered statistically significant.

RESULTS

Median follow-up from primary surgical resection was 12.8 months (range: 0.1-162.8), and median age was 58 years (20-85). Most patients were male (n=63; 65%) and had grade 4 tumors (n=71; 73%). OS differed by grade with 24-month actuarial OS rates of 81% and 34% (P&lt; 0.0001) for grade 3 and 4 gliomas, respectively. The presence of M0 (HR 2.2; 95% CI 1.4-3.6; P=0.001), M1 (HR 1.8; 95%CI 1.1-2.9; P=0.01), and M2 macrophages (HR 1.9; 95%CI 1.2-3.2; P=0.007) predicted OS. No other ICI subtypes were predictive of OS. The presence of M0- and M2-polarized macrophages were more common in grade 4 compared to grade 3 gliomas 46% vs. 11% (P=0.002) and 69% vs. 31% (P=0.0007), respectively.

CONCLUSION

The increased presence of non-polarized or M2 TAMs within the glioma microenvironment was significantly associated with OS in HGG. Their presence may serve as unique stratification and a potential therapeutic target.

Abstract LB-359: Gene expression profile and tumor infiltrating lymphocytes in Hispanic women with breast cancer

Breast cancer (BC) is the most commonly diagnosed cancer among women and is currently classified into four molecular subtypes. These subtypes have been characterized by their distinct immune gene profiles, where only triple-negative (TN) and HER2+ subtypes are recognized for being tumor immunogenic cancers. The main objective of this pilot study is to understand whether the immune gene expression profile from breast tumors can be associated with tumor infiltrating lymphocytes (TILs) levels in Hispanic women. Breast tumors were obtained through the Puerto Rico Biobank at Ponce Health Sciences University and in collaboration with Moffitt Cancer Center. Pathology reports were also obtained to collect information about tumor characteristics and molecular subtype classification. Fresh frozen (FF) (n=40) and FFPE (n=27) breast tumors were cut into slide sections and stained with H&E to confirm the presence of tumor cells. The percentage of TILs was measured by staining additional tumor sections for CD3+ T cells. TILs levels were scored by analyzing the infiltration of CD3+ T cells and were further classified into: low (≤40%) or high (&gt;40%). After microdissection of the breast tumors by a pathologist, total RNA was extracted to assess the expression changes of 770 immune genes (Pancancer Immune Panel, NanoString Technologies). The RNA quality was tested to ensure that it met the required specification for genomic analysis. Data was analyzed using the “randomForest” R package followed by proportion tests of the clinical data using contingency table analysis. All the samples were infiltrating ductal carcinomas with BC subtype distribution as follows: luminal A (n=50), luminal B (n=8), HER2+ (n=4), and TN (n=5). No significant differences were found after comparing the proportions of TILs between FF and FFPE tumor samples (p=1.00). Differentially expressed genes (DEGs) were obtained separately for FF and FFPE tumor with high and low TIL levels. The random forest analyses showed high predictive performance (~80% accuracy) to classify samples stratified by TIL levels in terms of these DEGs expressions. Sixteen (16) DEGs were found consistenly in FF and FFPE samples when stratifiying by TIL levels. The expression profiles of these DEGs show consistently higher expression in samples with high TILs levels. Among those DEGs are CLTA4 (p&lt;0.01, adj: FF 0.2030, FF: 0.0206), CXCL9 (p&lt;0.01, adj: FF: 0.203, FFPE: 0.229), CCL17 (p&lt;0.01, adj: FF: 0.151, FFPE: 0.203), and TIGIT (p&lt;0.01, adj: FF: 0.157, FFPE: 0.203). CLTA4 codifies for an immune checkpoint receptor while CXCL9 and CCL17 are known enhancers of antitumor immunity in BC. Another immune checkpoint molecule, was recently reported to be produced by BC cells to suppress the TH1 response. Our results suggest that the tumor cells can activate transcription factors that can affect the immune response in Hispanic women. Supported by grant: #4U54CA163071-05.

Citation Format: Jarline Encarnacion, Carmen Ortiz-Sanchez, Julie Dutil, Wandaliz Torres-Gracia, Steven Eschrich, Shari Pilon-Thomas, Jaime L. Matta. Gene expression profile and tumor infiltrating lymphocytes in Hispanic women with breast cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-359.

Abstract 3226: Facilitating personalized medicine with cloud-based storage and analytics

The vast wealth of medical data collected over the last decade holds great promise for accelerating novel research, discovery, and clinical translation. Specifically, the rapid expansion of genomic testing provides new opportunities for the clinical management of cancer patients, influencing diagnosis, risk stratification, and treatment planning. Moffitt Cancer Center's Personalized Medicine Clinical Service integrates next-generation sequencing test results into patient care, using the data to guide individualized treatment plans. To maximize the efficiency and efficacy of this service, creative solutions for data harmonization, storage, and management are required. We implemented a commercial molecular data warehouse (MDW), directly linked to our existing clinical data warehouse, to store and manage molecular data ranging from genotypic alterations to annotations from public resources (HUGO, COSMIC, Ensembl) and clinically actionable targets (4,256 records currently loaded). A centralized, cloud-based data and analytics platform is also being implemented at Moffitt that will integrate a broad range of multi-modal data. In the cloud environment, the data from the MDW will be linked to typically siloed data streams from the electronic health record, cancer registry management system, biospecimen management system, billing and scheduling systems, patient-reported information and outcomes, and patient-generated health data, creating a unique and customized Personalized Medicine Curated Data Mart (CDM). In addition to describing the features of the MDW and the challenges faced during its implementation, we will provide an overview of the extensive data cleaning and curation required to facilitate such a CDM. This includes the extraction of disease characteristics from unstructured clinical text via natural language processing, creation of new derived data fields, approaches to extracting and managing complex treatment data, and the inclusion of detailed, manually-abstracted recurrence and outcomes data for historical patients from existing institutional datasets such as the Clinical Genomics Action Committee (CGAC) database. Finally, we will present prototypes of analytics dashboards that will interface directly with our CDM, facilitating intuitive data exploration for all members of our personalized medicine teams.

Citation Format: Rachel Howard, Kevin Hicks, Jamie Teer, Phillip Reisman, Mandy O'Leary, Steven Eschrich, Ross Mitchell, Howard McLeod, Dana Rollison. Facilitating personalized medicine with cloud-based storage and analytics [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3226.

Empirically-derived synthetic populations to mitigate small sample sizes

Limited sample sizes can lead to spurious modeling findings in biomedical research. The objective of this work is to present a new method to generate synthetic populations (SPs) from limited samples using matched case-control data (n = 180 pairs), considered as two separate limited samples. SPs were generated with multivariate kernel density estimations (KDEs) with unconstrained bandwidth matrices. We included four continuous variables and one categorical variable for each individual. Bandwidth matrices were determined with Differential Evolution (DE) optimization by covariance comparisons. Four synthetic samples (n = 180) were derived from their respective SPs. Similarity between observed samples with synthetic samples was compared assuming their empirical probability density functions (EPDFs) were similar. EPDFs were compared with the maximum mean discrepancy (MMD) test statistic based on the Kernel Two-Sample Test. To evaluate similarity within a modeling context, EPDFs derived from the Principal Component Analysis (PCA) scores and residuals were summarized with the distance to the model in X-space (DModX) as additional comparisons. Four SPs were generated from each sample. The probability of selecting a replicate when randomly constructing synthetic samples (n = 180) was infinitesimally small. MMD tests indicated that the observed sample EPDFs were similar to the respective synthetic EPDFs. For the samples, PCA scores and residuals did not deviate significantly when compared with their respective synthetic samples. The feasibility of this approach was demonstrated by producing synthetic data at the individual level, statistically similar to the observed samples. The methodology coupled KDE with DE optimization and deployed novel similarity metrics derived from PCA. This approach could be used to generate larger-sized synthetic samples. To develop this approach into a research tool for data exploration purposes, additional evaluation with increased dimensionality is required. Moreover, given a fully specified population, the degree to which individuals can be discarded while synthesizing the respective population accurately will be investigated. When these objectives are addressed, comparisons with other techniques such as bootstrapping will be required for a complete evaluation.

Expression of CAS/CSE1L, the Cellular Apoptosis Susceptibility Protein, Correlates With Neoplastic Progression in Barrett's Esophagus

BACKGROUND

Identifying the molecular switch responsible for the neoplastic progression of Barrett's esophagus (BE) and initiation of adenocarcinoma (ADC) is clinically essential and it will have a profound impact on patient diagnosis, prognosis, and treatment. The cellular apoptosis susceptibility gene CAS/CSE1L is overexpressed in various cancers, including a rare report on esophageal ADC; however, its expression in BE neoplasia has not been addressed.

MATERIALS AND METHODS

We investigated the expression of the CAS/CSE1L protein immunohistochemically in 56 esophageal resection specimens for ADC arising in BE. For each specimen, a full representative section of the invasive ADC was selected to include, when possible, BE, low-grade dysplasia (LGD) and high-grade dysplasia (HGD). Samples were stained for CAS/CSE1L expression using a rabbit polyclonal antibody recognizing the N-terminus of human CAS/CSE1L. Protein expression levels were measured using the Allred semiquantitative scoring system. The data were evaluated using χ statistical analysis. Gene expression Omnibus was queried for CAS/CSE1L and BE neoplasia.

RESULTS

We found minimal to absent CAS/CSE1L in all BE tissue samples; however, CAS/CSE1L was upregulated in 60% of LGD and overexpressed in HGD and ADC. The results were statistically significant (P<0.05). The localization of CAS/CSE1L protein was nuclear in BE; it became nuclear and cytoplasmic in LGD and HGD, and predominantly cytoplasmic in ADC. A similar progressive increase was observed for CAS/CSE1L gene expression.

CONCLUSION

These findings show changes in CAS/CSE1L during BE progression. CAS/CSE1L may represent a potential marker for dysplasia/carcinoma.

Inhibition of the FAD containing ER oxidoreductin 1 (Ero1) protein by EN-460 as a strategy for treatment of multiple myeloma

Multiple myeloma (MM) cells demonstrate high basal endoplasmic reticulum (ER) stress and are typically exquisitely sensitive to agents such as proteasome inhibitors that activate the unfolded protein response. The flavin adenosine dinucleotide (FAD) containing endoplasmic reticulum oxidoreductin enzyme (Ero1L) catalyzes de-novo disulfide bridge formation of ER resident proteins and contributes to proper protein folding. Here we show that increased Ero1L expression is prognostic of poor outcomes for MM patients relapsing on therapy. We propose that targeting protein folding via inhibition of Ero1L may represent a novel therapeutic strategy for the treatment of MM. In this report we show that treatment of MM cells with EN-460, a known inhibitor of ERO1L, was sufficient to inhibit cell proliferation and induce apoptosis. Furthermore, we show that cell death correlated in part with induction of ER stress. We also show that EN460 inhibited the enzyme activity of Ero1L, with an IC50 of 22.13 μM, consistent with previous reports. However, EN-460 was also found to inhibit other FAD-containing enzymes including MAO-A (IC50 = 7.91 μM), MAO-B (IC50 = 30.59 μM) and LSD1 (IC50 = 4.16 μM), suggesting overlap in inhibitor activity and the potential need to develop more specific inhibitors to enable pharmacological validation of ERO1L as a target for the treatment of MM. We additionally prepared and characterized azide-tagged derivatives of EN-460 as possible functional probe compounds (e.g., for photo-affinity labeling) for future target-engagement studies and further development of structure-activity relationships.

Abstract 1565: OnPLS-based integrative proteogenomics analysis of lung squamous cell cancer

Introduction: Multivariate projection methods such as PCA and PLS has been widely applied for analysis of biological and chemical data. OnPLS is a recent extension to these methods suitable for integrative analysis of omics data. With OnPLS it is possible to compare multiple omics datasets to identify joint variation and variation locally unique for each of the studied datasets. OnPLS is a new approach for truly integrative analysis of omics data to be contrasted to commonly applied approaches limiting analysis to 1) comparing findings from individually analyzed blocks of data 2) pairwise comparison of individual probes.

Experimental: A Java based implementation of OnPLS was used for the statistical modeling. 116 lung squamous cell cancer samples were characterized using gene expression profiling and global proteomics. The OnPLS model was applied to jointly model variation between mRNA and protein expression. Enrichment analysis of factor loadings was performed using the Enrichr tools to identify biological mechanisms explained by the different joint and unique components of the OnPLS model.

Results: Using a cross-validation procedure the model with the highest predictive ability was calculated having two joint components and one locally unique component for each of the proteomics and gene expression datasets. The model explained 21.9% of the variation in the expression data and 26.1% of the variation in the proteomics data. The first joint component captures the highest degree of common variation between mRNA and protein activity. From the mRNA data, this component is related to immune infiltrates, especially monocytes and B-cells, whereas this component is related to extracellular matrix activity from the protein data. This suggests covariance of mRNA immune-related gene expression and extracellular matrix-related protein expression. As expected, local variation specific to the protein measurements involved regulation of protein activation and processing. mRNA-specific variation is related to keratinization, a key process in squamous cell cancer.

Conclusion: OnPLS offers an interesting approach for integrative analysis of omics data. Applying this approach to proteo-genomics data of lung squamous cell cancers suggest similar patterns of activity is represented in protein and gene expression data, however the biological processes associated with this activity may be distinct.

Citation Format: Fredrik Pettersson, Paul A. Stewart, Robbert J. Slebos, Eric A. Welsh, Ling Cen, Yonghong Zhang, Zhihua Chen, Chia-Ho Cheng, Guolin Zhang, Bin Fang, Victoria Izumi, Sean Yoder, Katherine Fellows, Y Ann Chen, Jamie K. Teer, Steven Eschrich, John M. Koomen, Anders Berglund, Eric B. Haura. OnPLS-based integrative proteogenomics analysis of lung squamous cell cancer [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 1565. doi:10.1158/1538-7445.AM2017-1565

Abstract 206: Proteogenomic classifications and outcome in squamous cell carcinoma of the lung

Introduction: Genomic analyses have yielded a tremendous amount of knowledge on the genetic changes in lung cancers, but translating this information into actionable data benefitting patients has proven difficult. The integration of proteomic analyses with genomics and gene expression profiling allows a more detailed description of the biological processes, thus improving our understanding of cancer phenotypes. These insights can potentially be used for better classification and help to guide patient selection for targeted therapies.

Experimental: We analyzed 116 surgically resected squamous cell lung carcinoma samples for copy-number alterations, gene expression profiling, targeted exome-sequencing and global proteomic profiling. The cohort consisted of mostly early stage tumors (83% Stage I or II) with complete follow-up (median 58 months). Copy number status was analyzed using the Affymetrix CytoScan array, DNA mutation status was assessed using a customized version of the Agilent Comprehensive Cancer Panel for targeted sequencing, and gene expression profiling was carried out by RNA-sequencing. Proteomic analysis was performed using TMT labeling, 12- fraction bRPLC separation and LC-MS/MS analysis with a Thermo Q-Exactive mass spectrometer. Database searches were performed using multiple search engines against RefSeq version 78, and summarized using IDPicker 3.

Results: The non-redundant protein inventory consisted of more than 6,000 protein groups with a protein FDR &lt;5%. Tumors were classified according to well established gene expression criteria into 4 classes: Classical, Basal, Primitive and Secretory. At the protein level, the Classical subtype was associated with xenobiotic and energy metabolism; the Basal subtype with defense responses and extracellular matrix changes; the Primitive subtype with nucleic acid metabolism; and the Secretory subtype with p38 signaling. These findings confirm and expand on previous mRNA expression studies of squamous cell lung carcinoma. Proteomics-based classification identified two sub-categories within the Classical subtype, which were characterized by inflammatory and stress response signaling. Within this group, patients with high expression of inflammation-associated proteins had better cancer-specific survival than those with low expression (p=0.04, Log-rank test). Targeted exome sequencing of 154 cancer-associated genes revealed frequent mutations in TP53, CDKN2A, NFE2L2, and other genes. Proteomic expression of genes located in amplified chromosomal regions was used to identify driver genes in squamous cell lung carcinoma.

Conclusion: Our results provide new biological insights from the addition of protein measurements to genomic datasets that have the potential to improve classification. The data suggest that proteins involved in immune responses are important for the biological behavior and outcome of the Classical subtype in squamous cell lung carcinoma.

Citation Format: Robbert J.C. Slebos, Paul A. Stewart, Eric A. Welsh, Ling Cen, Yonghong Zhang, Zhihua Chen, Chia-Ho Cheng, Fredrik Pettersson, Anders E. Berglund, Guolin Zhang, Bin Fang, Victoria Izumi, Sean J. Yoder, Katherine M. Fellows, Y Ann Chen, Jamie Teer, Steven Eschrich, John Koomen, Eric Haura. Proteogenomic classifications and outcome in squamous cell carcinoma of the lung [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 206. doi:10.1158/1538-7445.AM2017-206

The radiosensitivity index predicts for overall survival in glioblastoma

We have previously developed a multigene expression model of tumor radiosensitivity (RSI) with clinical validation in multiple cohorts and disease sites. We hypothesized RSI would identify glioblastoma patients who would respond to radiation and predict treatment outcomes. Clinical and array based gene expression (Affymetrix HT Human Genome U133 Array Plate Set) level 2 data was downloaded from the cancer genome atlas (TCGA). A total of 270 patients were identified for the analysis: 214 who underwent radiotherapy and temozolomide and 56 who did not undergo radiotherapy. Median follow-up for the entire cohort was 9.1 months (range: 0.04–92.2 months). Patients who did not receive radiotherapy were more likely to be older (p < 0.001) and of poorer performance status (p < 0.001). On multivariate analysis, RSI is an independent predictor of OS (HR = 1.64, 95% CI 1.08–2.5; p = 0.02). Furthermore, on subset analysis, radiosensitive patients had significantly improved OS in the patients with high MGMT expression (unmethylated MGMT), 1 year OS 84.1% vs. 53.7% (p = 0.005). This observation held on MVA (HR = 1.94, 95% CI 1.19–3.31; p = 0.008), suggesting that RT has a larger therapeutic impact in these patients. In conclusion, RSI predicts for OS in glioblastoma. These data further confirm the value of RSI as a disease-site independent biomarker.

Abstract LB-140: Scaling discovery proteomics to large lung cancer cohorts using data independent acquisition

The success of label free protein quantification relies on larger cohort sizes to accurately profile protein expression patterns enabling specific targets to be translated to confirmatory studies. Thus, proteome profiling experiments must satisfy two primary requirements: maximizing breadth and depth of protein sampling as well as facilitating automated data processing. The approach presented here integrates discovery experiments used to create reference databases and rapid data independent acquisition (DIA) methods to examine large clinical cohorts using &lt; 3 hours of instrument time per individual patient's resected tumor specimen. The combination facilitates exhaustive, automated data mining for protein identification and relative quantitation.

Samples used to evaluate the robustness of the study were collected from lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD). A sample total of 159 (50 LUSC, 46 LUAD, and 63 adjacent control lung tissues) were analyzed using UHPLC separations coupled to a hybrid quadrupole-orbital ion trap mass spectrometer (QExactive Plus) for DIA. Automated data processing was performed by matching peptides in spectral libraries developed from expression proteomics experiments using pooled tissue proteomes.

Data were evaluated to determine differences between LUSC, LUAD, and adjacent control tissue, which were then compared to previous literature to prioritize candidate biomarkers and evaluate the performance of DIA LC-MS/MS for tumor proteome profiling. Hierarchical clustering visualized potential tumor classification schemes based on proteomic phenotypes. In a further study, analysis of cores from a tissue microarray produced quantitative data for &gt; 3,000 proteins per core, indicating that the technology can be applied to minimal amounts of formalin fixed paraffin embedded tumor tissue. Because TMAs are assembled to address specific clinical questions, further analysis of these cohorts is an extremely valuable use for DIA; tumor proteome profiles can be rapidly accumulated and mapped to the clinical variables used to select samples for the TMA.

Data independent acquisition provides a method for discovery proteomics that balances a sufficient depth of coverage with the ability to analyze large cohorts of patients (n = 100-400) within 1-2 months on a single instrument. Initial data have been produced for lung squamous cell carcinoma and lung adenocarcinoma, which indicate its utility for assessment of patient groups.

Citation Format: Scott Peterman, Bin Fang, Melissa Hoffmann, Amol Prakash, Paul A. Stewart, Richard Liu, Matthew Smith, Joseph Johnson, Steven Eschrich, Guolin Zhang, Eric Haura, John Koomen. Scaling discovery proteomics to large lung cancer cohorts using data independent acquisition. [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 LB-140.

Differential association of STK11 and TP53 with KRAS mutation-associated gene expression, proliferation and immune surveillance in lung adenocarcinoma

While mutations in the KRAS oncogene are among the most prevalent in human cancer, there are few successful treatments to target these tumors. It is also likely that heterogeneity in KRAS-mutant tumor biology significantly contributes to the response to therapy. We hypothesized that the presence of commonly co-occurring mutations in STK11 and TP53 tumor suppressors may represent a significant source of heterogeneity in KRAS-mutant tumors. To address this, we utilized a large cohort of resected tumors from 442 lung adenocarcinoma patients with data including annotation of prevalent driver mutations (KRAS and EGFR) and tumor suppressor mutations (STK11 and TP53), microarray-based gene expression and clinical covariates, including overall survival (OS). Specifically, we determined impact of STK11 and TP53 mutations on a new KRAS mutation-associated gene expression signature as well as previously defined signatures of tumor cell proliferation and immune surveillance responses. Interestingly, STK11, but not TP53 mutations, were associated with highly elevated expression of KRAS mutation-associated genes. Mutations in TP53 and STK11 also impacted tumor biology regardless of KRAS status, with TP53 strongly associated with enhanced proliferation and STK11 with suppression of immune surveillance. These findings illustrate the remarkably distinct ways through which tumor suppressor mutations may contribute to heterogeneity in KRAS-mutant tumor biology. In addition, these studies point to novel associations between gene mutations and immune surveillance that could impact the response to immunotherapy.

ZEB1 Mediates Acquired Resistance to the Epidermal Growth Factor Receptor-Tyrosine Kinase Inhibitors in Non-Small Cell Lung Cancer

Epithelial-mesenchymal transition (EMT) is one mechanism of acquired resistance to inhibitors of the epidermal growth factor receptor-tyrosine kinases (EGFR-TKIs) in non-small cell lung cancer (NSCLC). The precise mechanisms of EMT-related acquired resistance to EGFR-TKIs in NSCLC remain unclear. We generated erlotinib-resistant HCC4006 cells (HCC4006ER) by chronic exposure of EGFR-mutant HCC4006 cells to increasing concentrations of erlotinib. HCC4006ER cells acquired an EMT phenotype and activation of the TGF-β/SMAD pathway, while lacking both T790M secondary EGFR mutation and MET gene amplification. We employed gene expression microarrays in HCC4006 and HCC4006ER cells to better understand the mechanism of acquired EGFR-TKI resistance with EMT. At the mRNA level, ZEB1 (TCF8), a known regulator of EMT, was >20-fold higher in HCC4006ER cells than in HCC4006 cells, and increased ZEB1 protein level was also detected. Furthermore, numerous ZEB1 responsive genes, such as CDH1 (E-cadherin), ST14, and vimentin, were coordinately regulated along with increased ZEB1 in HCC4006ER cells. We also identified ZEB1 overexpression and an EMT phenotype in several NSCLC cells and human NSCLC samples with acquired EGFR-TKI resistance. Short-interfering RNA against ZEB1 reversed the EMT phenotype and, importantly, restored erlotinib sensitivity in HCC4006ER cells. The level of micro-RNA-200c, which can negatively regulate ZEB1, was significantly reduced in HCC4006ER cells. Our results suggest that increased ZEB1 can drive EMT-related acquired resistance to EGFR-TKIs in NSCLC. Attempts should be made to explore targeting ZEB1 to resensitize TKI-resistant tumors.

Abstract 3752: Integrating proteomics and metabolomics characterizes active pathways and potential drug targets in small cell lung cancer

Small cell lung cancer (SCLC) accounts for 13-15% of all new lung cancer cases and represents the sixth most commonly diagnosed cancer in the US. However, it is an understudied cancer for which no molecularly targeted approaches have shown benefit. Using high-throughput techniques such as proteomics and metabolomics we can provide a more global, unbiased characterization to the inner workings of this disease. Integrating disparate data is a challenge in data analysis and bioinformatics, but this challenge is worth confronting due to the potential for combined proteomics / metabolomics analyses to better interrogate and capture the global landscape of active pathways and networks in SCLC than either technology alone. We accomplished this by comparing SCLC cell lines with non-small cell lung cancer (NSCLC) cell lines, differentiating characteristics of these two lung cancer cell types. We first used activity-based protein profiling (ABPP) combined with LC-MS/MS to profile the ATP-binding proteome of SCLC cell lines (n = 18) and NSCLC cell lines (n = 18). ABPP uses chemical probes that are directed against the active sites of enzymes to interrogate the functional state of ATP-binding enzymes, particularly kinases, in biological samples. These experiments identified 6937 peptides (2319 proteins), of which 3891 peptides (1543 proteins) were differentially expressed. Several pathways related to metabolism, such as purine biosynthesis and glycolysis / gluconeogenesis, were identified as over-represented in this list. These results led us to perform broad spectrum UPLC-TOF-MS metabolomics on ten SCLC and ten NSCLC cell lines. Multivariate analysis demonstrated distinct metabolite profiles for SCLC and NSCLC. Over 100 metabolites with variable importance to projection greater than 1 contributed to the differentiation of the two groups. These included metabolites related to purine metabolism such as inosinic acid and adenosine monophosphate and suggest a connection between our proteomics and metabolomics results. Statistical modeling approaches such as linear modeling, nonparametric nonlinear correlation, and Bayesian network analysis were used to integrate proteomics and metabolomics data to jointly characterize the key pathways and constituent components in SCLC. Multiple regression revealed a statistically significant interaction between endoglin (ENG) and phosphatidylethanolamine (PE), both of which are involved in angiogenesis. These findings and other statistical modeling results have the potential to facilitate the identification of new subtypes in SCLC and allow for the identification of novel targeted therapies. Future work will use these results to help characterize SCLC patient-derived xenograft models and surgically resected patient tissue samples. Funded, in part, by NIH Common Fund, 1U24DK097193.

Citation Format: Paul A. Stewart, Jiannong Li, Kate J. Fisher, Suraj Dhungana, Delisha Stewart, Susan Sumner, Eric Gardner, John Poirier, Charles M. Rudin, Eric A. Welsh, Steven Eschrich, Ann Chen, Eric B. Haura. Integrating proteomics and metabolomics characterizes active pathways and potential drug targets in small cell lung cancer. [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 3752. doi:10.1158/1538-7445.AM2015-3752

Tolerance associated gene expression following allogeneic hematopoietic cell transplantation

Biologic markers of immune tolerance may facilitate tailoring of immune suppression duration after allogeneic hematopoietic cell transplantation (HCT). In a cross-sectional study, peripheral blood samples were obtained from tolerant (n = 15, median 38.5 months post-HCT) and non-tolerant (n = 17, median 39.5 post-HCT) HCT recipients and healthy control subjects (n = 10) for analysis of immune cell subsets and differential gene expression. There were no significant differences in immune subsets across groups. We identified 281 probe sets unique to the tolerant (TOL) group and 122 for non-tolerant (non-TOL). These were enriched for process networks including NK cell cytotoxicity, antigen presentation, lymphocyte proliferation, and cell cycle and apoptosis. Differential gene expression was enriched for CD56, CD66, and CD14 human lineage-specific gene expression. Differential expression of 20 probe sets between groups was sufficient to develop a classifier with > 90% accuracy, correctly classifying 14/15 TOL cases and 15/17 non-TOL cases. These data suggest that differential gene expression can be utilized to accurately classify tolerant patients following HCT. Prospective investigation of immune tolerance biologic markers is warranted.

Quantitative computed tomographic descriptors associate tumor shape complexity and intratumor heterogeneity with prognosis in lung adenocarcinoma

Two CT features were developed to quantitatively describe lung adenocarcinomas by scoring tumor shape complexity (feature 1: convexity) and intratumor density variation (feature 2: entropy ratio) in routinely obtained diagnostic CT scans. The developed quantitative features were analyzed in two independent cohorts (cohort 1: n = 61; cohort 2: n = 47) of patients diagnosed with primary lung adenocarcinoma, retrospectively curated to include imaging and clinical data. Preoperative chest CTs were segmented semi-automatically. Segmented tumor regions were further subdivided into core and boundary sub-regions, to quantify intensity variations across the tumor. Reproducibility of the features was evaluated in an independent test-retest dataset of 32 patients. The proposed metrics showed high degree of reproducibility in a repeated experiment (concordance, CCC≥0.897; dynamic range, DR≥0.92). Association with overall survival was evaluated by Cox proportional hazard regression, Kaplan-Meier survival curves, and the log-rank test. Both features were associated with overall survival (convexity: p = 0.008; entropy ratio: p = 0.04) in Cohort 1 but not in Cohort 2 (convexity: p = 0.7; entropy ratio: p = 0.8). In both cohorts, these features were found to be descriptive and demonstrated the link between imaging characteristics and patient survival in lung adenocarcinoma.

Epigenomic characterization of locally advanced anal cancer: a radiation therapy oncology group 98-11 specimen study

BACKGROUND

The Radiation Therapy Oncology Group 98-11 clinical trial demonstrated the superiority of standard 5-fluorouracil/mitomycin-C over 5-fluorouracil/cisplatin in combination with radiation in the treatment of anal squamous cell cancer. Tumor size (>5 cm) and lymph node metastases are associated with disease progression. There may be key molecular differences (eg, DNA methylation changes) in tumors at high risk for progression.

OBJECTIVE

The objectives of this study were to determine whether there are differences in DNA methylation at individual CpG sites and within genes among locally advanced anal cancers, with large tumor size and/or nodal involvement, compared with those that are less advanced.

DESIGN

This was a case-case study among 121 patients defined as high risk (tumor size >5 cm and/or nodal involvement; n = 59) or low risk (≤5 cm, node negative; n = 62) within the mitomycin-C arm of the Radiation Therapy Oncology Group 98-11 trial. DNA methylation was measured using the Illumina HumanMethylation450 Array.

SETTINGS

The study was conducted in a tertiary care cancer center in collaboration with a national clinical trials cooperative group.

PATIENTS

The patients consisted of 74 women and 47 men with a median age of 54 years (range, 25-79 years).

MAIN OUTCOME MEASURES

DNA methylation differences at individual CpG sites and within genes between low- and high-risk patients were compared using the Mann-Whitney test (p < 0.001).

RESULTS

A total of 16 CpG loci were differentially methylated (14 increased and 2 decreased) in high- versus low-risk cases. Genes harboring differentially methylated CpG sites included known tumor suppressor genes and novel targets.

LIMITATIONS

This study only included patients in the mitomycin-C arm with tumor tissue; however, this sample was representative of the trial.

CONCLUSIONS

This is the first study to apply genome-wide methylation analysis to anal cancer. Biologically relevant differences in methylated targets were found to discriminate locally advanced from early anal cancer. Epigenetic events likely play a significant role in the progression of anal cancer and may serve as potential biomarkers.

Perturbation of the mutated EGFR interactome identifies vulnerabilities and resistance mechanisms

We hypothesized that elucidating the interactome of epidermal growth factor receptor (EGFR) forms that are mutated in lung cancer, via global analysis of protein-protein interactions, phosphorylation, and systematically perturbing the ensuing network nodes, should offer a new, more systems-level perspective of the molecular etiology. Here, we describe an EGFR interactome of 263 proteins and offer a 14-protein core network critical to the viability of multiple EGFR-mutated lung cancer cells. Cells with acquired resistance to EGFR tyrosine kinase inhibitors (TKIs) had differential dependence of the core network proteins based on the underlying molecular mechanisms of resistance. Of the 14 proteins, 9 are shown to be specifically associated with survival of EGFR-mutated lung cancer cell lines. This included EGFR, GRB2, MK12, SHC1, ARAF, CD11B, ARHG5, GLU2B, and CD11A. With the use of a drug network associated with the core network proteins, we identified two compounds, midostaurin and lestaurtinib, that could overcome drug resistance through direct EGFR inhibition when combined with erlotinib. Our results, enabled by interactome mapping, suggest new targets and combination therapies that could circumvent EGFR TKI resistance.

Abstract PR08: Identification of novel methylated genomic regions in invasive cervical cancer using a genome-wide approach

Introduction: In the US, an estimated 35,000 cancer cases per year are attributable to Human papillomavirus (HPV), including cancers of the cervix, anus and oropharynx. Despite the availability of an HPV-targeted vaccine, issues with low uptake and significant lag time from vaccination to cancer prevention will result in cervical cancer continuing to be a substantial burden to US and world-wide populations. It is becoming increasingly clear that epigenetic events play a critical role in HPV-associated carcinogenesis, including changes in host genome DNA methylation patterns. In the cervix, methylation patterns have been shown to be altered; however, there have been inconclusive findings as to the most informative methylation targets in cervical cancer. We hypothesize that there are novel epigenetic alterations in HPV-associated cervical cancer and these changes may be early detection biomarkers in pre-invasive cervical lesions. In this study, we sought to identify epigenetic differences in HPV-positive cervical cancers compared to HPV-positive normal tissue using genome-wide methylation arrays and to explore if these differences can be identified in cervical intraepithelial neoplasia (CIN) 3 lesions.

Methods: DNA was extracted from 24 frozen cervical tissues (9 normal, 5 CIN 3 and 10 invasive cancers) following pathology review and macrodissection. Bisulfite-modified DNA was interrogated for methylation at over 450,000 CpG loci using the Illumina HumanMethylation450 BeadChip array. HPV genotyping was conducted using SFP10 LiPA HPV-typing system. Differential methylation between normal and tumors at individual CpG loci was determined by t-test with significance set by a false discovery rate (or q-value) and final sets were restricted to those with a mean difference in β-values≥0.3. Next, using a window-based approach, we examined clusters of CpG loci within gene coding regions or CpG islands. Finally, a binary classification strategy was used for each gene set of probes, where a probe was defined as methylated if the β-values were above the minimum point in the bi-modal histogram. Sensitivity and specificity for each gene at the set methylation cut-point was determined.

Results: A total of 24 HPV positive tissues were examined in this study (9 normal, 5 CIN 3 and 10 invasive cancers). A total of 321 CpG loci displayed differential methylation with mean difference ≥0.5 in tumors compared to normal (FDR corrected q&lt;0.0005), of which 310 were hypermethylated and 11 were hypomethylated in tumors. We examined CpG site clusters within gene-coding regions among a larger set of 13,515 loci differentially methylated using less stringent criteria (q&lt;0.01, β-value difference≥0.3). A total of 369 genes demonstrated differential methylation across clusters of CpG loci (12-1293 CpG sites per gene) between normal and tumors. Of these 369 genes, 80 were differentially methylated in ≥50% of CpG loci evaluated within the gene. Methylation status of 19 genes had both a sensitivity and specificity for distinguishing tumor from normal of 100%, including MIR129-2, protocadherin 10 (PCDH-10), and POU class 3 homeobox 3 (POU3F3). All 19 genes were also differentially methylated in at least one CIN 3 lesion.

Conclusions: We have identified several novel methylated CpG loci and genomic regions in HPV-associated cervical cancer, some of which were also found to be altered in CIN 3 lesions. Validation of these novel CpG loci in a larger population of women with cervical dysplasia and cancer is warranted. A number of these novel methylated sites may emerge as promising biomarkers for the early detection of cervical cancer and possibly, for other HPV-associated malignancies.

This abstract is also presented as Poster C24.

Citation Format: Erin M. Siegel, Anders Berglund, Bridget Riggs, Steven Eschrich, Maddox Kristen, Abidemi Ajidahun, David Shibata, Kevin D. Brown. Identification of novel methylated genomic regions in invasive cervical cancer using a genome-wide approach. [abstract]. In: Proceedings of the Twelfth Annual AACR International Conference on Frontiers in Cancer Prevention Research; 2013 Oct 27-30; National Harbor, MD. Philadelphia (PA): AACR; Can Prev Res 2013;6(11 Suppl): Abstract nr PR08.

Development and application of a novel metric to assess effectiveness of biomedical data

OBJECTIVE

Design a metric to assess the comparative effectiveness of biomedical data elements within a study that incorporates their statistical relatedness to a given outcome variable as well as a measurement of the quality of their underlying data.

MATERIALS AND METHODS

The cohort consisted of 874 patients with adenocarcinoma of the lung, each with 47 clinical data elements. The p value for each element was calculated using the Cox proportional hazard univariable regression model with overall survival as the endpoint. An attribute or A-score was calculated by quantification of an element's four quality attributes; Completeness, Comprehensiveness, Consistency and Overall-cost. An effectiveness or E-score was obtained by calculating the conditional probabilities of the p-value and A-score within the given data set with their product equaling the effectiveness score (E-score).

RESULTS

The E-score metric provided information about the utility of an element beyond an outcome-related p value ranking. E-scores for elements age-at-diagnosis, gender and tobacco-use showed utility above what their respective p values alone would indicate due to their relative ease of acquisition, that is, higher A-scores. Conversely, elements surgery-site, histologic-type and pathological-TNM stage were down-ranked in comparison to their p values based on lower A-scores caused by significantly higher acquisition costs.

CONCLUSIONS

A novel metric termed E-score was developed which incorporates standard statistics with data quality metrics and was tested on elements from a large lung cohort. Results show that an element's underlying data quality is an important consideration in addition to p value correlation to outcome when determining the element's clinical or research utility in a study.

Abstract 937: Dissection of KRAS-driven survival signaling networks via phosphoproteomics in lung cancer cells

KRAS mutation accounts for development of approximately 30% of lung adenocarcinoma yet therapeutic options remain limited. We report our efforts to explore downstream signaling proteins driven by KRAS, including TANK-binding kinase 1 (TBK1), through a mass spectrometry based phosphoproteomics approach. We hypothesized that such a search would identify key survival proteins as well as uncover potential adaptive resistance mechanisms. We identified proteins whose phosphorylation is regulated TBK1 using SILAC and mass spectrometry following RNAi-mediated TBK1 knockdown. A cohesive network view of the underlying results centered major effects on Polo-like Kinase 1 (PLK1) and decreased phosphorylation of its targets implicating TBK1 as involved in mitosis. TBK1 activity was induced especially in late G1 and M phase, and loss of TBK1 sensitized lung cancer cells to a mitotic stressor. Surprisingly, we found that TBK1 knockdown increased phosphorylation of oncogenic kinases, including EGFR, Met, and ERK1/2. We screened multiple tyrosine kinase inhibitors in combination with TBK1 loss and found the pro-apoptotic effect of TBK1 loss was enhanced by dasatinib, a Src-family kinase inhibitor, providing a mechanistic basis for rational combinatorial therapies involving TBK1. Moving upstream, we analyzed global phosphoproteome change after KRAS knockdown and identified unexpected role of KRAS involved in CDK1 regulation as well as potential adaptive resistant mechanisms that can protect cells against KRAS loss. Finally, we have started exploring changes in the programmed kinome in KRAS mutated lung cancer cells following exposure to MEK inhibitors using activity based protein profiling. Initial experiments have characterized phenotypic effects of two MEK inhibitors in a subset of KRAS mutated lung cancer cells and preliminary studies have identified increased levels of some kinases following MEK inhibition. Collectively, we expect this kinome and phosphoproteomic network based approach will provide better insights into survival signaling mechanism maintaining survival of KRAS mutant lung cancer cell as well as novel therapeutic strategies for this cancer subtype. Updated work will be presented. Supported by the SPORE in Lung Cancer (P50-CA119997)

Citation Format: Jae-Young Kim, Eric A. Welsh, Bin Fang, Umut Oguz, Jiannong Li, Fumi Kinose, Crystina Bronk, Amer A. Beg, Ann Chen, Steven Eschrich, John Koomen, Eric B. Haura. Dissection of KRAS-driven survival signaling networks via phosphoproteomics in lung cancer cells. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 937. doi:10.1158/1538-7445.AM2013-937

42. Epigenomic profiling of anal cancer: does size matter? An RTOG 98-11 specimen study

Background Anal cancer accounts for 4% of all lower gastrointestinal tract malignancies in the US. One of the most important predictors of prognosis among anal cancer patients is the size of the primary tumour. Tumours with diameters >5 cm have poorer disease-free survival than those with smaller tumours. Understanding the biological changes associated with tumour growth may provide information to guide therapy and improve patient outcomes. DNA methylation changes are critical epigenetic events in cancer development. Methods: In this study we sought to characterise the epigenomic signatures associated with anal cancer tumour size (≤5 cm vs >5 cm) in FFPE tissues from 121 patients (≤5 cm = 88; >5 cm = 33) who participated in the RTOG 98–11 cooperative group anal cancer clinical trial. Differential methylation, examined at >450 000 CpG loci using the Illumina Human Methylation 450 Array, were compared between the two groups using Mann–Whitney test (significance = P < 0.001 and difference in methylation β-value >0.1). Results: This study included 74 women and 47 men with a median age of 54 years. A total of 86 CpG loci were differentially methylated (78 increased and 8 decreased) in large vs small tumours. Genes harbouring CpG sites that were among the most highly differentially methylated included those associated with WNT signalling (FZD10, WNT9A), microRNAs (MIR200A) and novel methylated targets (PON3). Conclusions: These data provide evidence that epigenetic events likely play a significant role in the progression of anal SCC and may serve as biomarkers of prognosis. Similar epigenomic approaches may be useful at earlier stages of anal neoplastic progression for application in screening and early detection.

DNA methylation profiling across the spectrum of HPV-associated anal squamous neoplasia

BACKGROUND

Changes in host tumor genome DNA methylation patterns are among the molecular alterations associated with HPV-related carcinogenesis. However, there is little known about the epigenetic changes associated specifically with the development of anal squamous cell cancer (SCC). We sought to characterize broad methylation profiles across the spectrum of anal squamous neoplasia.

METHODOLOGY/PRINCIPAL FINDINGS

Twenty-nine formalin-fixed paraffin embedded samples from 24 patients were evaluated and included adjacent histologically normal anal mucosa (NM; n = 3), SCC-in situ (SCC-IS; n = 11) and invasive SCC (n = 15). Thirteen women and 11 men with a median age of 44 years (range 26-81) were included in the study. Using the SFP(10) LiPA HPV-typing system, HPV was detected in at least one tissue from all patients with 93% (27/29) being positive for high-risk HPV types and 14 (93%) of 15 invasive SCC tissues testing positive for HPV 16. Bisulfite-modified DNA was interrogated for methylation at 1,505 CpG loci representing 807 genes using the Illumina GoldenGate Methylation Array. When comparing the progression from normal anal mucosa and SCC-IS to invasive SCC, 22 CpG loci representing 20 genes demonstrated significant differential methylation (p<0.01). The majority of differentially methylated gene targets occurred at or close to specific chromosomal locations such as previously described HPV methylation "hotspots" and viral integration sites.

CONCLUSIONS

We have identified a panel of differentially methlylated CpG loci across the spectrum of HPV-associated squamous neoplasia of the anus. To our knowledge, this is the first reported application of large-scale high throughput methylation analysis for the study of anal neoplasia. Our findings support further investigations into the role of host-genome methylation in HPV-associated anal carcinogenesis with implications towards enhanced diagnosis and screening strategies.

12-Chemokine gene signature identifies lymph node-like structures in melanoma: potential for patient selection for immunotherapy?

We have interrogated a 12-chemokine gene expression signature (GES) on genomic arrays of 14,492 distinct solid tumors and show broad distribution across different histologies. We hypothesized that this 12-chemokine GES might accurately predict a unique intratumoral immune reaction in stage IV (non-locoregional) melanoma metastases. The 12-chemokine GES predicted the presence of unique, lymph node-like structures, containing CD20⁺ B cell follicles with prominent areas of CD3⁺ T cells (both CD4⁺ and CD8⁺ subsets). CD86⁺, but not FoxP3⁺, cells were present within these unique structures as well. The direct correlation between the 12-chemokine GES score and the presence of unique, lymph nodal structures was also associated with better overall survival of the subset of melanoma patients. The use of this novel 12-chemokine GES may reveal basic information on in situ mechanisms of the anti-tumor immune response, potentially leading to improvements in the identification and selection of melanoma patients most suitable for immunotherapy.

The metabolomic signature of malignant glioma reflects accelerated anabolic metabolism

Although considerable progress has been made toward understanding glioblastoma biology through large-scale genetic and protein expression analyses, little is known about the underlying metabolic alterations promoting their aggressive phenotype. We conducted global metabolomic profiling on patient-derived glioma specimens and identified specific metabolic programs differentiating low- and high-grade tumors, with the metabolic signature of glioblastoma reflecting accelerated anabolic metabolism. When coupled with transcriptional profiles, we identified the metabolic phenotype of the mesenchymal subtype to consist of accumulation of the glycolytic intermediate phosphoenolpyruvate and decreased pyruvate kinase activity. Unbiased hierarchical clustering of metabolomic profiles identified three subclasses, which we term energetic, anabolic, and phospholipid catabolism with prognostic relevance. These studies represent the first global metabolomic profiling of glioma, offering a previously undescribed window into their metabolic heterogeneity, and provide the requisite framework for strategies designed to target metabolism in this rapidly fatal malignancy.

Abstract 3395: TGFβ response signature in non-small cell lung carcinoma

Background: Transforming growth factor beta (TGFβ) is a cytokine involved in numerous cellular processes that include proliferation, migration and apoptosis. TGFβ is intriguing in that it plays a dual role in cancer; in some cases it functions as a tumor suppressor while in others it acts as a tumor promoter. Genetic alterations in the TGFβ signaling pathway have been reported in many cancers including breast, colon, gastric, lung, head and neck and pancreatic cancers. Correlations between a TGFβ induced gene expression and clinical outcomes have been described in breast and liver cancers, but not in non-small cell lung cancer (NSCLC). Because TGFβ has been shown to induce an epithelial-to-mesenchymal transition (EMT) in NSCLC that may lead to increased potential to invade and disseminate, we hypothesized that a TGFβ-induced gene expression signature might correlate with this transition and might predict prognosis in NSCLC patients. We therefore decided to attempt to identify gene expression changes induced by TGFβ that correlated with induced proliferation, migration, and EMT. We further investigated whether these TGFβ induced genes might be useful in the classification of NSCLC tumors. Methods: Ten different NSCLC cell lines were assessed for the ability of TGFβ-1 to induce migration, proliferation, and EMT. Cell lines were treated with TGFβ-1 for various time periods and the RNA collected for microarray analysis. Results: In nine out of the ten cell lines the TGFβ pathway was activated after treatment with TGFβ-1, but these lines differ in their TGFβ proliferation, migration and EMT responses. The H1944, H358, and A549 cell lines respond by shifting from an epithelial to a mesenchymal phenotype. Using the before and after treatment samples of these cell lines we looked for genes correlated with this transition. Initially, 632 probesets (534 genes) were identified that changed similarly in these 3 cell lines. Many of these changes seen after 120 hours of treatment were also seen at 48 hours but to a lesser degree. After Principal Component Analysis of the 534 genes in tumor sample a core set of 70 genes were retained and used to analyze lung cancer tumor samples. Using two independently derived sets of lung tumors the EMT signature could be used to identify 3 classes of tumors: tumors with high EMT gene expression, tumors with low EMT expression, and a distinct group of tumors that do not coordinately express the EMT genes identified in this study. Conclusion: Analysis of the differential responses of NSCLC cell lines to TGFβ-1 allows us to group NSCLC cell lines into at least three groups based on their response to TGFβ treatment. Using an EMT signature developed from the cell lines it appears that tumor samples can also be classified into three or more distinct groups that may differ in their in vivo responses to TGFβ-1.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3395. doi:1538-7445.AM2012-3395

Phosphoproteomics identifies driver tyrosine kinases in sarcoma cell lines and tumors

Driver tyrosine kinase mutations are rare in sarcomas, and patterns of tyrosine phosphorylation are poorly understood. To better understand the signaling pathways active in sarcoma, we examined global tyrosine phosphorylation in sarcoma cell lines and human tumor samples. Anti-phosphotyrosine antibodies were used to purify tyrosine phosphorylated peptides, which were then identified by liquid chromatography and tandem mass spectrometry. The findings were validated with RNA interference, rescue, and small-molecule tyrosine kinase inhibitors. We identified 1,936 unique tyrosine phosphorylated peptides, corresponding to 844 unique phosphotyrosine proteins. In sarcoma cells alone, peptides corresponding to 39 tyrosine kinases were found. Four of 10 cell lines showed dependence on tyrosine kinases for growth and/or survival, including platelet-derived growth factor receptor (PDGFR)α, MET, insulin receptor/insulin-like growth factor receptor signaling, and SRC family kinase signaling. Rhabdomyosarcoma samples showed overexpression of PDGFRα in 13% of examined cases, and sarcomas showed abundant tyrosine phosphorylation and expression of a number of tyrosine phosphorylated tyrosine kinases, including DDR2, EphB4, TYR2, AXL, SRC, LYN, and FAK. Together, our findings suggest that integrating global phosphoproteomics with functional analyses with kinase inhibitors can identify drivers of sarcoma growth and survival.

Abstract IA2: Network models in oncogene-addicted lung cancer

Lung cancer is a devastating worldwide disease yet enthusiasm exists for treatment of subsets of the disease with molecularly targeted agents. Mutations in the epidermal growth factor receptor (EGFR) or translocation of echinoderm microtubule associated protein like 4 – anaplastic lymphoma kinase (EML4-ALK) define two unique subsets of lung cancer characterized by sensitivity to tyrosine kinase inhibitors (TKI). Despite striking results with TKI, not all patients respond, the drugs are non-curative, and resistance is universal. Mutations in KRAS also define a group of patients awaiting therapeutic opportunities. We are characterizing signaling networks using tandem affinity purification (TAP) and liquid chromatography-mass spectrometry (LC-MS/MS) to map protein-protein interactions (PPI) and anti-phosphotyrosine immunoprecipitation coupled with LC-MS/MS to map tyrosine phosphorylation. In PC9 cells with mutated EGFR, we characterized a physical EGFR network consisting of 266 proteins by integrating both TAP and pTyr MS data. In H3122 cells harboring EML4-ALK, we identified a PPI network consisting of 113 proteins and using pTyr MS identified changes in tyrosine phosphorylation in 120 proteins (58 decreased, 62 increased) following exposure to ALK tyrosine kinase inhibitor. Functional proteins are being discovered from these networks using siRNA and inhibitor screens. In parallel studies, we have exploited the use of chemical proteomics to discern targets of promiscuous kinase inhibitors and enable optimal combination approaches. In this way, we view network mapping linked to chemical proteomics as one approach to discern novel drug combination studies for in vivo validation and ultimately translation to early phase clinical trials. To translate PPI based mass spectrometry studies to clinic, we are developing in situ assays that identify and quantify PPI using proximal ligation assays (PLA). Pilot studies identify EGFR:Grb2 interactions in formalin fixed human lung cancers. These approaches have the potential to enable ‘network medicine’ by identifying novel combination approaches as well as through identifying subtypes of cancers through network views of cancer.

Is Error-Based Pruning Redeemable?

Error based pruning can be used to prune a decision tree and it does not require the use of validation data. It is implemented in the widely used C4.5 decision tree software. It uses a parameter, the certainty factor, that affects the size of the pruned tree. Several researchers have compared error based pruning with other approaches, and have shown results that suggest that error based pruning results in larger trees that give no increase in accuracy. They further suggest that as more data is added to the training set, the tree size after applying error based pruning continues to grow even though there is no increase in accuracy. It appears that these results were obtained with the default certainty factor value. Here, we show that varying the certainty factor allows significantly smaller trees to be obtained with minimal or no accuracy loss. Also, the growth of tree size with added data can be halted with an appropriate choice of certainty factor. Methods of determining the certainty factor are discussed for both small and large data sets. Experimental results support the conclusion that error based pruning can be used to produce appropriately sized trees with good accuracy when compared with reduced error pruning.

Necdin, a negative growth regulator, is a novel STAT3 target gene down-regulated in human cancer

Cytokine and growth factor signaling pathways involving STAT3 are frequently constitutively activated in many human primary tumors, and are known for the transcriptional role they play in controlling cell growth and cell cycle progression. However, the extent of STAT3's reach on transcriptional control of the genome as a whole remains an important question. We predicted that this persistent STAT3 signaling affects a wide variety of cellular functions, many of which still remain to be characterized. We took a broad approach to identify novel STAT3 regulated genes by examining changes in the genome-wide gene expression profile by microarray, using cells expressing constitutively-activated STAT3. Using computational analysis, we were able to define the gene expression profiles of cells containing activated STAT3 and identify candidate target genes with a wide range of biological functions. Among these genes we identified Necdin, a negative growth regulator, as a novel STAT3 target gene, whose expression is down-regulated at the mRNA and protein levels when STAT3 is constitutively active. This repression is STAT3 dependent, since inhibition of STAT3 using siRNA restores Necdin expression. A STAT3 DNA-binding site was identified in the Necdin promoter and both EMSA and chromatin immunoprecipitation confirm binding of STAT3 to this region. Necdin expression has previously been shown to be down-regulated in a melanoma and a drug-resistant ovarian cancer cell line. Further analysis of Necdin expression demonstrated repression in a STAT3-dependent manner in human melanoma, prostate and breast cancer cell lines. These results suggest that STAT3 coordinates expression of genes involved in multiple metabolic and biosynthetic pathways, integrating signals that lead to global transcriptional changes and oncogenesis. STAT3 may exert its oncogenic effect by up-regulating transcription of genes involved in promoting growth and proliferation, but also by down-regulating expression of negative regulators of the same cellular processes, such as Necdin.

Abstract ED01-02: Imaging ‘omics in cancer detection

“Radiomics” is the process of extracting quantitative descriptive features from biomedical images in high throughput. These image features are thus mineable data that can be used to parse against medical outcomes or gene expression data. A long term goal is to develop models using these data that will enable improved detection, diagnosis and therapy prediction. Two projects in lung cancer are currently underway.

The first focuses on non-small cell lung cancer (NSCLC) imaged with standard-of-care PET/CT and CT images with 3–5mm resolution. For CT scans, the Radiomic dataset includes 219 three-dimensional and 102 two-dimensional features from each CT images. Data have been generated for over 1,000 lung cancer patients who have undergone surgical biopsy. Of these, 276 samples have been analyzed for gene expression using Affymetrix platforms and analyses are underway to determine relationships between image features, clinical outcomes and gene expression. In developing this technology, we have improved methods for automated delineation of lung lesions. This uses a multi-seed point approach to initiate region growing algorithm to generate an ensemble segmentation from multiple regions. This has been evaluated on a set of lung tumor datasets from the Moffitt Cancer Center thoracic oncology database. The similarity index (SI) was calculated as the ratio of the union over the intersection across multiple segmentations from the same lesion. In this metric, a SI of 100% reports identical results from every initial seed. For our data set, the average SI was &gt; 93% in 121/129 patients with 20 different start seed points for each case. This data set has been uploaded to CABIG using the NBIA platform and is available as a “segmentation challenge”. In our own data set, if PET scans are available, we are using centroids of the maximal SUV to automatically detect seed points. Another important step in the qualification process is to assess the quality of the data through co-variance, test-retest reproducibility and biological ranges of these features. The full feature sets have been analyzed for covariance, and a number of redundant sets of features have been identified, which allowed a reduction in the dimensionality of feature space. Test-retest analyses have baseline and follow-up thoracic CT scans (slice thickness 1.25mm) that were obtained within 15 minutes on 32 patients with NSCLC. All patients had primary pulmonary tumor of 1 cm or larger. Following segmentation, 190 3D quantitative features were extracted and the reproducibility of these tumor features in the two scans performed on each patient was assessed by calculating the concordance correlation coefficient (CCC). In general, the intra-patient reproducibility of all features was high, 102/190 showed excellent reproducibility (CCC&gt;0.75). Notably, the inter-patient biological ranges for individual features were highly variable. Additionally, a co-variance matrix of features identified several redundancies in the feature set which could be combined into a single variable. Combining inter-scan variance, biological range and co-variance, we have reduced the total number of features from 190 to a set of ∼20 that may be the most informative. Comparisons of quantitative imaging feature data with patient outcome have begun with an approach to classify Adenocarcinoma, Squamous-cell Carcinoma and Bronchioalveolar Carcinoma (BAC). Classifiers including decision trees and support vector machines are used along with feature selection techniques (Wrappers and Relief-F) to build effective models for tumor classification. Results show that over the large feature space for 3D features it is possible to recognize tumor classes with over 76% accuracy, showing new features may be of help. The results indicate that image features on their own can be used to identify tumor classes with relatively high accuracy.

The second project has just been initiated to use the Radiomics approach to extract features from screening CTs. This is using the ACRIN-sponsored National Lung Screening Trial, NLST, database of ca. 50,000 individuals who were screened with CT or planar X-ray. The results of this study are now being published and have shown a significant survival benefit in the CT-screened population. A concern with this approach is the large number of false positives as well as nagging false negatives, i.e. patients who presented with late-stage cancer following a “negative” exam. The Radiomics approach will test the hypothesis that quantitative texture and other features will add to predictive power and result in improved sensitivity and specificity. Because the data set is large, significant effort is being expended to automate the 3-D segmentations and identify regions of interest within the lung fields.

Citation Information: Cancer Prev Res 2011;4(10 Suppl):ED01-02.

BVES regulates EMT in human corneal and colon cancer cells and is silenced via promoter methylation in human colorectal carcinoma

The acquisition of a mesenchymal phenotype is a critical step in the metastatic progression of epithelial carcinomas. Adherens junctions (AJs) are required for suppressing this epithelial-mesenchymal transition (EMT) but less is known about the role of tight junctions (TJs) in this process. Here, we investigated the functions of blood vessel epicardial substance (BVES, also known as POPDC1 and POP1), an integral membrane protein that regulates TJ formation. BVES was found to be underexpressed in all stages of human colorectal carcinoma (CRC) and in adenomatous polyps, indicating its suppression occurs early in transformation. Similarly, the majority of CRC cell lines tested exhibited decreased BVES expression and promoter DNA hypermethylation, a modification associated with transcriptional silencing. Treatment with a DNA-demethylating agent restored BVES expression in CRC cell lines, indicating that methylation represses BVES expression. Reexpression of BVES in CRC cell lines promoted an epithelial phenotype, featuring decreased proliferation, migration, invasion, and anchorage-independent growth; impaired growth of an orthotopic xenograft; and blocked metastasis. Conversely, interfering with BVES function by expressing a dominant-negative mutant in human corneal epithelial cells induced mesenchymal features. These biological outcomes were associated with changes in AJ and TJ composition and related signaling. Therefore, BVES prevents EMT, and its epigenetic silencing may be an important step in promoting EMT programs during colon carcinogenesis.

BVES regulates EMT in human corneal and colon cancer cells and is silenced via promoter methylation in human colorectal carcinoma

The acquisition of a mesenchymal phenotype is a critical step in the metastatic progression of epithelial carcinomas. Adherens junctions (AJs) are required for suppressing this epithelial-mesenchymal transition (EMT) but less is known about the role of tight junctions (TJs) in this process. Here, we investigated the functions of blood vessel epicardial substance (BVES, also known as POPDC1 and POP1), an integral membrane protein that regulates TJ formation. BVES was found to be underexpressed in all stages of human colorectal carcinoma (CRC) and in adenomatous polyps, indicating its suppression occurs early in transformation. Similarly, the majority of CRC cell lines tested exhibited decreased BVES expression and promoter DNA hypermethylation, a modification associated with transcriptional silencing. Treatment with a DNA-demethylating agent restored BVES expression in CRC cell lines, indicating that methylation represses BVES expression. Reexpression of BVES in CRC cell lines promoted an epithelial phenotype, featuring decreased proliferation, migration, invasion, and anchorage-independent growth; impaired growth of an orthotopic xenograft; and blocked metastasis. Conversely, interfering with BVES function by expressing a dominant-negative mutant in human corneal epithelial cells induced mesenchymal features. These biological outcomes were associated with changes in AJ and TJ composition and related signaling. Therefore, BVES prevents EMT, and its epigenetic silencing may be an important step in promoting EMT programs during colon carcinogenesis.

Abstract 1627: Proteome-wide analysis of echinoderm microtubule associated protein like 4 – anaplastic lymphoma kinase (EML4-ALK) network in lung cancer

The translocation of echinoderm microtubule associated protein like 4 – anaplastic lymphoma kinase (EML4-ALK) defines a unique, new subset of patients with non small cell lung cancer (NSCLC). It represents a novel diagnostic biomarker and therapeutic target for NSCLC. PF 02341066 is a ALK small tyrosine kinase inhibitor that has demonstrated high clinical activity in EML4-ALK- positive NSCLC patients. We sought to characterize the signaling network used by EML4-ALK and reveal potential new therapeutic targets. We used tandem affinity purification (TAP) and liquid chromatography-mass spectrometry (LC-MS/MS) to map protein-protein interactions and anti-phosphotyrosine immunoprecipitation coupled with LC-MS/MS to map tyrosine phosphorylation. We selected the human NSCLC H3122 cell line for our studies as it is sensitive to PF 02341066 and harbors ELM4-ALK. Quantitative western blotting results show that activities of MET, ALK, AKT, ERK and STAT3 were inhibited by treatment with PF 02341066. Proteome-wide analysis of tyrosine phosphorylation of the ALK signaling pathway were examined. In total, 311 tyrosine phosphorylated proteins (487 unique phosphorylated sites) were identified. Of those, 73 pTyr sites (62proteins) were upregulated and 78 pTyr sites (58 proteins) were downregulated by 1uM PF02341066. Using TAP, we identified 13 proteints interacting with EML4-ALK including GRB2 and SHC1. Core interactions were verified by coupling of immunoprecipitation to Western blot and mass spectrometry. A comprehensive physical ALK network was generated by integrating both phosphoproteomic data and interactional partners using pathway analysis algorithms. Two adaptors, SHC1and GAB2, are suggested as the first outlets of signaling propagation from ALK. Other sublevels directly connected with ALK may allow further understanding the mechanism of ‘ALK addiction’. A sub-network composed of 35 cell death associated proteins was created. An ongoing functional analysis including siRNA and small inhibitor screen on this small network should validate druggable targets. This network approach has potential translational applications to identify novel approaches to treatment of ALK addicted lung cancers and could be useful in patients with acquired resistance to ALK inhibitors.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1627. doi:10.1158/1538-7445.AM2011-1627

Abstract 5111: Integrative phospho-proteomic and genomic analyses identify AXL as a potential biomarker and therapeutic target for NRAS-mutated melanoma

Currently, few therapeutic options exist for the 15-20% of patients whose melanomas harbor activating mutations in NRAS. Here, we have used comprehensive phospho-proteomic and genomic analysis to characterize the patterns of intracellular signaling in a panel of NRAS and BRAF-mutated melanoma cell lines, with the goal of identifying new biomarkers and therapeutic targets for NRAS mutated melanomas. Our preliminary analyses showed BRAF mutated melanomas to have less diverse intracellular signaling than the NRAS group, which tended to have constitutive phosphorylation in a wide-range of receptor tyrosine kinases (RTKs), such as c-MET, EGFR, HER2 and c-Abl. Although these patterns of RTK activity were heterogeneous across the NRAS mutated melanoma cell line panel, all of the NRAS mutated cell lines examined showed constitutive activity in the receptor tyrosine kinase Axl. A role for Axl in the oncogenic behavior of NRAS mutated melanomas was suggested by array CGH studies showing genomic amplification of Axl in a significant fraction of the NRAS mutated melanoma cell lines and by Western Blot experiments demonstrating increased expression of Axl and its ligand Gas6. Mass spectrometry analysis and immunoprecipitation revealed Axl to be constitutively phosphorylated at the Tyr702 autophosphorylation site in NRAS mutated melanoma cell lines only. Functionally, siRNA knockdown and pharmacological inhibition of Axl impaired the survival and invasion of NRAS mutated melanoma cell lines grown under 3D organotypic cell culture conditions. Studies are ongoing to confirm the role of Axl in the initiation and progression of melanomas harboring NRAS mutations and determine whether Axl constitutes a therapeutic target for this melanoma sub-group.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 5111. doi:10.1158/1538-7445.AM2011-5111

Abstract 1919: Baseline characteristics of colorectal cancer patients enrolled in the colorectal cancer outcomes, prognosis and epidemiology cohort (The COPE) study

Background: Obesity is a public health problem and an increasing number of colorectal cancer patients will be obese. The impact of obesity on the efficacy of colorectal cancer treatment and patient outcomes has only recently been elucidated. However, evidence is lacking on the mechanism linking obesity to poor outcomes among colorectal cancer patients. There are emerging data to suggest that the underlying mechanism for this association is mediated in part by hyperinsulinemia and insulin resistance, via the insulin-insulin-like growth factor (IGF) axis, and modulated in part by adiponectin. Using a multidisciplinary approach, the Colorectal Cancer Outcomes, Prognosis and Epidemiology Study will elucidate the biological effects of elevated insulin and IGF-factors on tumor gene expression and their impact on quality of life.

Objective: The objectives of this presentation are to present the study design and baseline characterize of colorectal cancer participants enrolled to date in the COPE study.

Methods: We are establishing a cohort of 350 colorectal cancer patients identified at the Moffitt Cancer Center. All patients will be followed at regular intervals to obtain data on health outcomes, health behaviors, symptoms, and quality of life, and blood draws will occur at (∼6 and 12 months post-enrollment). An electronic intake form collected use of dietary supplements and of medications (including multivitamins, folate, and NSAIDs, with type and frequency); smoking, and a brief assessment of physical activity at baseline. BMI, waist and hip measures are taken at each visit. The assessment of quality of life and symptoms will be based on validated standard instruments (European Organization for Research and Treatment of Cancer QLQ-C30) administered at baseline, 3, 6 and 12 month follow-up time points, with slight adaptations in timing for patients undergoing chemotherapy.

Results: Overall, a total of seventy five colorectal cancer patients have been enrolled. Among these patients, 60% have colon cancer and 40% were diagnosed with rectal cancer. A majority of the patients were married with some higher education. Sixty percent of the participants were female. There was an equal distribution of cases enrolled across stages I – III, with very few stage IV patients. A full characterization of the cohort will be presented. Conclusion: Overall, the COPE study is designed to recruit a hospital based cohort of colorectal cancer cases to evaluate the association between body size, insulin/IGF and colorectal cancer outcomes. We have successfully recruited and followed patients over a twelve month period. We are now poised to evaluate our primary objectives among a cohort of colorectal cancer patients.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1919. doi:10.1158/1538-7445.AM2011-1919