Samir Lababidi

Comparison of RNA-seq and microarray-based models for clinical endpoint prediction

BackgroundGene expression profiling is being widely applied in cancer research to identify biomarkers for clinical endpoint prediction. Since RNA-seq provides a powerful tool for transcriptome-based applications beyond the limitations of microarrays, we sought to systematically evaluate the performance of RNA-seq-based and microarray-based classifiers in this MAQC-III/SEQC study for clinical endpoint prediction using neuroblastoma as a model.ResultsWe generate gene expression profiles from 498 primary neuroblastomas using both RNA-seq and 44 k microarrays. Characterization of the neuroblastoma transcriptome by RNA-seq reveals that more than 48,000 genes and 200,000 transcripts are being expressed in this malignancy. We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays. To systematically compare the power of RNA-seq and microarray-based models in predicting clinical endpoints, we divide the cohort randomly into training and validation sets and develop 360 predictive models on six clinical endpoints of varying predictability. Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.ConclusionsWe demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction. Our findings may be valuable to guide future studies on the development of gene expression-based predictive models and their implementation in clinical practice.

Targeting Angiogenesis via a c-Myc/Hypoxia-Inducible Factor-1α–Dependent Pathway in Multiple Myeloma

Bone marrow angiogenesis is associated with multiple myeloma (MM) progression. Here, we report high constitutive hypoxia-inducible factor-1alpha (Hif-1alpha) expression in MM cells, which is associated with oncogenic c-Myc. A drug screen for anti-MM agents that decrease Hif-1alpha and c-Myc levels identified a variety of compounds, including bortezomib, lenalidomide, enzastaurin, and adaphostin. Functionally, based on transient knockdowns and overexpression, our data delineate a c-Myc/Hif-1alpha-dependent pathway mediating vascular endothelial growth factor production and secretion. The antiangiogenic activity of our tool compound, adaphostin, was subsequently shown in a zebrafish model and translated into a preclinical in vitro and in vivo model of MM in the bone marrow milieu. Our data, therefore, identify Hif-1alpha as a novel molecular target in MM and add another facet to anti-MM drug activity.

Gene markers of cellular aging in human multipotent stromal cells in culture

IntroductionHuman multipotent stromal cells (MSCs) isolated from bone marrow or other tissue sources have great potential to treat a wide range of injuries and disorders in the field of regenerative medicine and tissue engineering. In particular, MSCs have inherent characteristics to suppress the immune system and are being studied in clinical studies to prevent graft-versus-host disease. MSCs can be expanded in vitro and have potential for differentiation into multiple cell lineages. However, the impact of cell passaging on gene expression and function of the cells has not been determined.MethodsCommercially available human MSCs derived from bone marrow from six different donors, grown under identical culture conditions and harvested at cell passages 3, 5, and 7, were analyzed with gene-expression profiling by using microarray technology.ResultsThe phenotype of these cells did not change as reported previously; however, a statistical analysis revealed a set of 78 significant genes that were distinguishable in expression between passages 3 and 7. None of these significant genes corresponded to the markers established by the International Society for Cellular Therapy (ISCT) for MSC identification. When the significant gene lists were analyzed through pathway analysis, these genes were involved in the top-scoring networks of cellular growth and proliferation and cellular development. A meta-analysis of the literature for significant genes revealed that the MSCs seem to be undergoing differentiation into a senescent cell type when cultured extensively. Consistent with the differences in gene expression at passage 3 and 7, MSCs exhibited a significantly greater potential for cell division at passage 3 in comparison to passage 7.ConclusionsOur results identified specific gene markers that distinguish aging MSCs grown in cell culture. Confirmatory studies are needed to correlate these molecular markers with biologic attributes that may facilitate the development of assays to test the quality of MSCs before clinical use.

Interleukin‐4 receptor alpha overexpression in human bladder cancer correlates with the pathological grade and stage of the disease

Previously, we have demonstrated that interleukin‐4 receptor α (IL‐4Rα) is overexpressed on a variety of human cancers and can serve as target for IL‐4 immunotoxin comprised of IL‐4 and a mutated Pseudomonas exotoxin. However, its expression and association with grade and clinical stage of bladder cancer has not been studied. IL‐4Rα expression was examined in human bladder cancer cell lines, mouse xenografts, and biopsy specimens at mRNA and protein levels by real‐time RT‐PCR and IHC/ISH techniques. We also examined the effect of IL‐4 on proliferation and invasion of bladder carcinoma cell lines. For tissue microarray (TMA) results, we analyzed the precision data using exact binomial proportion with exact two‐sided P‐values. We used Cochran–Armitage Statistics with exact two‐sided P‐values to examine the trend analysis of IL‐4Rα over grade or stage of the bladder cancer specimens. The influence of age and gender covariates was also analyzed using multiple logistic regression models. IL‐4Rα is overexpressed in five bladder cancer cell lines, while normal bladder and human umbilical vein cell lines (HUVEC) expressed at low levels. Two other chains of IL‐4 receptor complex, IL‐2RγC and IL‐13Rα1, were absent or weakly expressed. IL‐4 modestly inhibited the cell proliferation, but enhanced cell invasion of bladder cancer cell lines in a concentration‐dependent manner. Bladder cancer xenografts in immunodeficient mice also maintained IL‐4Rα overexpression in vivo. Analysis of tumor biopsy specimens in TMAs revealed significantly higher IL‐4Rα immunostaining (≥2+) in Grade 2 (85%) and Grade 3 (97%) compared to Grade 1 tumors (0%) (P ≤ 0.0001). Similarly, 9% stage I tumors were positive for IL‐4Rα (≥2+) compared to 84% stage II (P ≤ 0.0001) and 100% stages III–IV tumors (P ≤ 0.0001). IL‐13Rα1 was also expressed in tumor tissues but at low levels and it did not show any correlation with the grade and stage of disease. However, the IL‐2RγC was not expressed. Ten normal bladder specimens demonstrated ≤1+ staining for IL‐4Rα and IL‐13Rα1 and no staining for IL‐2RγC. These results demonstrate that IL‐4Rα is overexpressed in human bladder cancer, which correlates with advanced grade and stage of the disease. Thus, IL‐4Rα may be a bladder tumor‐associated protein and a prognostic biomarker.

Women participating in a dietary intervention trial maintain dietary changes without much effect on household members.

Abstract: This study examined whether subjects who participated in a 12-mo intervention would maintain their diets 1 yr after the study ended and whether the diets of household members were affected. Premenopausal women, who had at least one first-degree relative with breast cancer (n = 122), were randomized to one of four diets: control, low fat (15% of energy), high fruit and vegetable (FV, nine servings per day), and combination low fat, high FV. Study subjects and one household member were asked to complete the Block ‘95 food-frequency questionnaire (FFQ) at baseline, 1 yr, and 2 yr. Study subjects also completed 24-h recalls and 4-day food records at baseline and Year 1. Fat and FV intakes by all three assessment methods compared reasonably well except that fat intakes by FFQ were somewhat higher. FV intakes by FFQ in the high-FV and combination arms increased significantly from 4 servings per day to about 10 servings per day at Year 1 and 7 servings per day at Year 2. FV intakes increased much more modestly in the low-fat and control arms. Fat intakes in the low-fat and combination arms were lower at Year 1 than Year 2, but mean Year 2 fat intakes of 26–28% were still significantly lower than those at baseline. In household members, the only significant change was a small decrease in energy from fat at Year 1 in the household members of subjects who were in the combination arm. These results indicate that study subjects were making large dietary changes independently of their household members and that fat and FV intakes in study subjects 1 yr after intervention stopped were still substantially different from intakes at baseline.

In silico analysis of autoimmune diseases and genetic relationships to vaccination against infectious diseases.

BackgroundNear universal administration of vaccines mandates intense pharmacovigilance for vaccine safety and a stringently low tolerance for adverse events. Reports of autoimmune diseases (AID) following vaccination have been challenging to evaluate given the high rates of vaccination, background incidence of autoimmunity, and low incidence and variable times for onset of AID after vaccinations. In order to identify biologically plausible pathways to adverse autoimmune events of vaccine-related AID, we used a systems biology approach to create a matrix of innate and adaptive immune mechanisms active in specific diseases, responses to vaccine antigens, adjuvants, preservatives and stabilizers, for the most common vaccine-associated AID found in the Vaccine Adverse Event Reporting System.ResultsThis report focuses on Guillain-Barre Syndrome (GBS), Rheumatoid Arthritis (RA), Systemic Lupus Erythematosus (SLE), and Idiopathic (or immune) Thrombocytopenic Purpura (ITP). Multiple curated databases and automated text mining of PubMed literature identified 667 genes associated with RA, 448 with SLE, 49 with ITP and 73 with GBS. While all data sources provided valuable and unique gene associations, text mining using natural language processing (NLP) algorithms provided the most information but required curation to remove incorrect associations. Six genes were associated with all four AIDs. Thirty-three pathways were shared by the four AIDs. Classification of genes into twelve immune system related categories identified more “Th17 T-cell subtype” genes in RA than the other AIDs, and more “Chemokine plus Receptors” genes associated with RA than SLE. Gene networks were visualized and clustered into interconnected modules with specific gene clusters for each AID, including one in RA with ten C-X-C motif chemokines. The intersection of genes associated with GBS, GBS peptide auto-antigens, influenza A infection, and influenza vaccination created a subnetwork of genes that inferred a possible role for the MAPK signaling pathway in influenza vaccine related GBS.ConclusionsResults showing unique and common gene sets, pathways, immune system categories and functional clusters of genes in four autoimmune diseases suggest it is possible to develop molecular classifications of autoimmune and inflammatory events. Combining this information with cellular and other disease responses should greatly aid in the assessment of potential immune-mediated adverse events following vaccination.

Uncertainty estimation with a finite dataset in the assessment of classification models

To successfully translate genomic classifiers to the clinical practice, it is essential to obtain reliable and reproducible measurement of the classifier performance. A point estimate of the classifier performance has to be accompanied with a measure of its uncertainty. In general, this uncertainty arises from both the finite size of the training set and the finite size of the testing set. The training variability is a measure of classifier stability and is particularly important when the training sample size is small. Methods have been developed for estimating such variability for the performance metric AUC (area under the ROC curve) under two paradigms: a smoothed cross-validation paradigm and an independent validation paradigm. The methodology is demonstrated on three clinical microarray datasets in the microarray quality control consortium phase two project (MAQC-II): breast cancer, multiple myeloma, and neuroblastoma. The results show that the classifier performance is associated with large variability and the estimated performance may change dramatically on different datasets. Moreover, the training variability is found to be of the same order as the testing variability for the datasets and models considered. In conclusion, the feasibility of quantifying both training and testing variability of classifier performance is demonstrated on finite real-world datasets. The large variability of the performance estimates shows that patient sample size is still the bottleneck of the microarray problem and the training variability is not negligible.

Challenges in DNA Microarray Studies from the Regulatory Perspective

Genomic classifiers using DNA microarrays are becoming powerful tools in the medical community with the potential to revolutionize the diagnosis and treatment of disease. However, despite the tremendous interest in using these classifiers in diagnosis and the management of disease, few genomic classifiers have made it into clinical practice. Some of the major challenges for the development and validation of genomic classifiers will be discussed in this article together with some of their difficulties.

Overall conceptual framework for studying the genetics of autoimmune diseases following vaccination: a regulatory perspective

The US Vaccine Adverse Event Reporting System contains case reports of autoimmune diseases (ADs) occurring following vaccinations. ADs are rare and occur in unvaccinated people, making the potential association between vaccines and ADs challenging to evaluate. Developing mechanistic pathways that link genes, immune mediators, vaccine components and ADs would be helpful for hypothesis generation, enhancing theories of biologic plausibility and grouping rare autoimmune adverse events to increase the ability to detect and evaluate safety signals. Here, we propose a conceptual framework for investigating the genetics of ADs as safety signals following vaccination, potentially contributing to the identification of relevant biomarkers. We also discuss a study design that incorporates genetic information into postmarket clinical evaluation of autoimmune adverse events following vaccination.