Leslie Cope

Pre-processing Agilent microarray data.

BackgroundPre-processing methods for two-sample long oligonucleotide arrays, specifically the Agilent technology, have not been extensively studied. The goal of this study is to quantify some of the sources of error that affect measurement of expression using Agilent arrays and to compare Agilents Feature Extraction software with pre-processing methods that have become the standard for normalization of cDNA arrays. These include log transformation followed by loess normalization with or without background subtraction and often a between array scale normalization procedure. The larger goal is to define best study design and pre-processing practices for Agilent arrays, and we offer some suggestions.ResultsSimple loess normalization without background subtraction produced the lowest variability. However, without background subtraction, fold changes were biased towards zero, particularly at low intensities. ROC analysis of a spike-in experiment showed that differentially expressed genes are most reliably detected when background is not subtracted. Loess normalization and no background subtraction yielded an AUC of 99.7% compared with 88.8% for Agilent processed fold changes. All methods performed well when error was taken into account by t- or z-statistics, AUCs ≥ 99.8%. A substantial proportion of genes showed dye effects, 43% (99%CI : 39%, 47%). However, these effects were generally small regardless of the pre-processing method.ConclusionSimple loess normalization without background subtraction resulted in low variance fold changes that more reliably ranked gene expression than the other methods. While t-statistics and other measures that take variation into account, including Agilents z-statistic, can also be used to reliably select differentially expressed genes, fold changes are a standard measure of differential expression for exploratory work, cross platform comparison, and biological interpretation and can not be entirely replaced. Although dye effects are small for most genes, many array features are affected. Therefore, an experimental design that incorporates dye swaps or a common reference could be valuable.

Modeling precision treatment of breast cancer

BackgroundFirst-generation molecular profiles for human breast cancers have enabled the identification of features that can predict therapeutic response; however, little is known about how the various data types can best be combined to yield optimal predictors. Collections of breast cancer cell lines mirror many aspects of breast cancer molecular pathobiology, and measurements of their omic and biological therapeutic responses are well-suited for development of strategies to identify the most predictive molecular feature sets.ResultsWe used least squares-support vector machines and random forest algorithms to identify molecular features associated with responses of a collection of 70 breast cancer cell lines to 90 experimental or approved therapeutic agents. The datasets analyzed included measurements of copy number aberrations, mutations, gene and isoform expression, promoter methylation and protein expression. Transcriptional subtype contributed strongly to response predictors for 25% of compounds, and adding other molecular data types improved prediction for 65%. No single molecular dataset consistently out-performed the others, suggesting that therapeutic response is mediated at multiple levels in the genome. Response predictors were developed and applied to TCGA data, and were found to be present in subsets of those patient samples.ConclusionsThese results suggest that matching patients to treatments based on transcriptional subtype will improve response rates, and inclusion of additional features from other profiling data types may provide additional benefit. Further, we suggest a systems biology strategy for guiding clinical trials so that patient cohorts most likely to respond to new therapies may be more efficiently identified.

Elevated microRNA miR-21 Levels in Pancreatic Cyst Fluid Are Predictive of Mucinous Precursor Lesions of Ductal Adenocarcinoma

Background: Biomarkers for the diagnostic classification of pancreatic cysts are urgently needed. Deregulated microRNA (miRNAs) expression is widespread in pancreatic cancer. We assessed whether aberrant miRNAs in pancreatic cyst fluid could be used as potential biomarkers for cystic precursor lesions of pancreatic cancer. Methods: Cyst fluid specimens were prospectively collected from 40 surgically resected pancreatic cysts, and small RNAs were extracted. The ‘mucinous’ cohort included 14 intraductal papillary mucinous neoplasms (including 3 with an associated adenocarcinoma) and 10 mucinous cystic neoplasms; the ‘nonmucinous’ cohort included 11 serous cystadenomas and 5 other benign cysts. Quantitative reverse transcription PCR was performed for five miRNAs (miR-21, miR-155, miR-221, miR-17-3p, miR-191), which were previously reported as overexpressed in pancreatic adenocarcinomas. Results: Significantly higher expression of miR-21, miR-221, and miR-17-3p was observed in the mucinous versus nonmucinous cysts (p < 0.01), with the mean relative fold differences being 7.0-, 7.9-, and 5.4-fold, respectively. Receiver operating characteristic curves demonstrated the highest median area under the curve for miR-21, with a median specificity of 76%, at a sensitivity of 80%. Conclusion: This pilot study demonstrates that profiling miRNAs in pancreatic cyst fluid samples is feasible and can yield potential biomarkers for the classification of cystic lesions of the pancreas.

Exploring Massive, Genome Scale Datasets with the GenometriCorr Package

We have created a statistically grounded tool for determining the correlation of genomewide data with other datasets or known biological features, intended to guide biological exploration of high-dimensional datasets, rather than providing immediate answers. The software enables several biologically motivated approaches to these data and here we describe the rationale and implementation for each approach. Our models and statistics are implemented in an R package that efficiently calculates the spatial correlation between two sets of genomic intervals (data and/or annotated features), for use as a metric of functional interaction. The software handles any type of pointwise or interval data and instead of running analyses with predefined metrics, it computes the significance and direction of several types of spatial association; this is intended to suggest potentially relevant relationships between the datasets. Availability and implementation: The package, GenometriCorr, can be freely downloaded at http://genometricorr.sourceforge.net/. Installation guidelines and examples are available from the sourceforge repository. The package is pending submission to Bioconductor.

Clinical Validation of KRAS, BRAF, and EGFR Mutation Detection Using Next-Generation Sequencing

OBJECTIVESnTo validate next-generation sequencing (NGS) technology for clinical diagnosis and to determine appropriate read depth.nnnMETHODSnWe validated the KRAS, BRAF, and EGFR genes within the Ion AmpliSeq Cancer Hotspot Panel using the Ion Torrent Personal Genome Machine (Life Technologies, Carlsbad, CA).nnnRESULTSnWe developed a statistical model to determine the read depth needed for a given percent tumor cellularity and number of functional genomes. Bottlenecking can result from too few input genomes. By using 16 formalin-fixed, paraffin-embedded (FFPE) cancer-free specimens and 118 cancer specimens with known mutation status, we validated the six traditional analytic performance characteristics recommended by the Next-Generation Sequencing: Standardization of Clinical Testing Working Group. Baseline noise is consistent with spontaneous and FFPE-induced C:G→T:A deamination mutations.nnnCONCLUSIONSnRedundant bioinformatic pipelines are essential, since a single analysis pipeline gave false-negative and false-positive results. NGS is sufficiently robust for the clinical detection of gene mutations, with attention to potential artifacts.

HMGA1 Reprograms Somatic Cells into Pluripotent Stem Cells by Inducing Stem Cell Transcriptional Networks

Background Although recent studies have identified genes expressed in human embryonic stem cells (hESCs) that induce pluripotency, the molecular underpinnings of normal stem cell function remain poorly understood. The high mobility group A1 (HMGA1) gene is highly expressed in hESCs and poorly differentiated, stem-like cancers; however, its role in these settings has been unclear. Methods/Principal Findings We show that HMGA1 is highly expressed in fully reprogrammed iPSCs and hESCs, with intermediate levels in ECCs and low levels in fibroblasts. When hESCs are induced to differentiate, HMGA1 decreases and parallels that of other pluripotency factors. Conversely, forced expression of HMGA1 blocks differentiation of hESCs. We also discovered that HMGA1 enhances cellular reprogramming of somatic cells to iPSCs together with the Yamanaka factors (OCT4, SOX2, KLF4, cMYC – OSKM). HMGA1 increases the number and size of iPSC colonies compared to OSKM controls. Surprisingly, there was normal differentiation in vitro and benign teratoma formation in vivo of the HMGA1-derived iPSCs. During the reprogramming process, HMGA1 induces the expression of pluripotency genes, including SOX2, LIN28, and cMYC, while knockdown of HMGA1 in hESCs results in the repression of these genes. Chromatin immunoprecipitation shows that HMGA1 binds to the promoters of these pluripotency genes in vivo. In addition, interfering with HMGA1 function using a short hairpin RNA or a dominant-negative construct blocks cellular reprogramming to a pluripotent state. Conclusions Our findings demonstrate for the first time that HMGA1 enhances cellular reprogramming from a somatic cell to a fully pluripotent stem cell. These findings identify a novel role for HMGA1 as a key regulator of the stem cell state by inducing transcriptional networks that drive pluripotency. Although further studies are needed, these HMGA1 pathways could be exploited in regenerative medicine or as novel therapeutic targets for poorly differentiated, stem-like cancers.

Association of PD-1/PD-L axis expression with cytolytic activity, mutational load, and prognosis in melanoma and other solid tumors

Significance The Cancer Genome Atlas datasets were used in the current study to explore the relationship of programmed death ligand-1 (PD-L1) expression, a cytotoxic T-cell gene signature, and mutational load to each other, to immunoactive factors, such as programmed cell death protein-1 (PD-1), PD-L2, and other checkpoint molecules, and to survival across multiple solid tumor types. We found that PD-L2 expression is more closely related to an ongoing host immune response in certain tumor types than PD-L1. Notably, mutational load was not immediately related to inflammation in any tumor type studied, and was inferior to an inflamed tumor microenvironment for predicting survival in patients with metastatic melanoma. Our findings also indicate the need for biomarker assays that are tumor-type–specific and include both expression studies and genomic profiling. Programmed cell death protein-1 (PD-1)/programmed death ligand-1 (PD-L1) checkpoint blockade has led to remarkable and durable objective responses in a number of different tumor types. A better understanding of factors associated with the PD-1/PD-L axis expression is desirable, as it informs their potential role as prognostic and predictive biomarkers and may suggest rational treatment combinations. In the current study, we analyzed PD-L1, PD-L2, PD-1, and cytolytic activity (CYT) expression, as well as mutational density from melanoma and eight other solid tumor types using The Cancer Genome Atlas database. We found that in some tumor types, PD-L2 expression is more closely linked to Th1/IFNG expression and PD-1 and CD8 signaling than PD-L1. In contrast, mutational load was not correlated with a Th1/IFNG gene signature in any tumor type. PD-L1, PD-L2, PD-1, CYT expression, and mutational density are all positive prognostic features in melanoma, and conditional inference modeling revealed PD-1/CYT expression (i.e., an inflamed tumor microenvironment) as the most impactful feature, followed by mutational density. This study elucidates the highly interdependent nature of these parameters, and also indicates that future biomarkers for anti–PD-1/PD-L1 will benefit from tumor-type–specific, integrated, mRNA, protein, and genomic approaches.

Targeting DDX3 with a small molecule inhibitor for lung cancer therapy

Lung cancer is the most common malignancy worldwide and is a focus for developing targeted therapies due to its refractory nature to current treatment. We identified a RNA helicase, DDX3, which is overexpressed in many cancer types including lung cancer and is associated with lower survival in lung cancer patients. We designed a first‐in‐class small molecule inhibitor, RK‐33, which binds to DDX3 and abrogates its activity. Inhibition of DDX3 by RK‐33 caused G1 cell cycle arrest, induced apoptosis, and promoted radiation sensitization in DDX3‐overexpressing cells. Importantly, RK‐33 in combination with radiation induced tumor regression in multiple mouse models of lung cancer. Mechanistically, loss of DDX3 function either by shRNA or by RK‐33 impaired Wnt signaling through disruption of the DDX3–β‐catenin axis and inhibited non‐homologous end joining—the major DNA repair pathway in mammalian somatic cells. Overall, inhibition of DDX3 by RK‐33 promotes tumor regression, thus providing a compelling argument to develop DDX3 inhibitors for lung cancer therapy.

Distinct DNA methylation profiles in ovarian serous neoplasms and their implications in ovarian carcinogenesis

OBJECTIVEnThe purpose of this study was to analyze DNA methylation profiles among different types of ovarian serous neoplasm, which is a task that has not been performed.nnnSTUDY DESIGNnThe Illumina beads array (Illumina Inc, San Diego, CA) was used to profile DNA methylation in enriched tumor cells that had been isolated from 75 benign and malignant serous tumor tissues and 6 tumor-associated stromal cell cultures.nnnRESULTSnWe found significantly fewer hypermethylated genes in high-grade serous carcinomas than in low-grade serous carcinoma and borderline tumors, which in turn had fewer hypermethylated genes than serous cystadenoma. Unsupervised analysis identified that serous cystadenoma, serous borderline tumor, and low-grade serous carcinomas tightly clustered together and were clearly different from high-grade serous carcinomas. We also performed supervised analysis to identify differentially methylated genes that may contribute to group separation.nnnCONCLUSIONnThe findings support the view that low-grade and high-grade serous carcinomas are distinctly different with low-grade, but not high-grade, serous carcinomas that are related to serous borderline tumor and cystadenoma.

Mutational analysis of BRAF and KRAS in ovarian serous borderline (atypical proliferative) tumours and associated peritoneal implants

There is debate as to whether peritoneal implants associated with serous borderline tumours/atypical proliferative serous tumours (SBT/APSTs) of the ovary are derived from the primary ovarian tumour or arise independently in the peritoneum. We analysed 57 SBT/APSTs from 45 patients with advanced‐stage disease identified from a nation‐wide tumour registry in Denmark. Mutational analysis for hotspots in KRAS and BRAF was successful in 55 APSTs and demonstrated KRAS mutations in 34 (61.8%) and BRAF mutations in eight (14.5%). Mutational analysis was successful in 56 peritoneal implants and revealed KRAS mutations in 34 (60.7%) and BRAF mutations in seven (12.5%). Mutational analysis could not be performed in two primary tumours and in nine implants, either because DNA amplification failed or because there was insufficient tissue for mutational analysis. For these specimens we performed VE1 immunohistochemistry, which was shown to be a specific and sensitive surrogate marker for a V600E BRAF mutation. VE1 staining was positive in one of two APSTs and seven of nine implants. Thus, among 63 implants for which mutation status was known (either by direct mutational analysis or by VE1 immunohistochemistry), 34 (53.9%) had KRAS mutations and 14 (22%) had BRAF mutations, of which identical KRAS mutations were found in 34 (91%) of 37 SBT/APST–implant pairs and identical BRAF mutations in 14 (100%) of 14 SBT/APST–implant pairs. Wild‐type KRAS and BRAF (at the loci investigated) were found in 11 (100%) of 11 SBT/APST–implant pairs. Overall concordance of KRAS and BRAF mutations was 95% in 59 of 62 SBT/APST–implant (non‐invasive and invasive) pairs (p < 0.00001). This study provides cogent evidence that the vast majority of peritoneal implants, non‐invasive and invasive, harbour the identical KRAS or BRAF mutations that are present in the associated SBT/APST, supporting the view that peritoneal implants are derived from the primary ovarian tumour. Copyright