Translating microarray data for diagnostic testing in childhood leukaemia

Respiratory health effects of residential individual and cumulative risk factors in children living in two cities of the Pearl River Delta Region, China

Background

Indoor environment is complex, with many factors potentially interacting with each other to affect health. However, previous studies have usually focused on effect of a single factor. Assessment of the combined effects of multiple factors can help with understanding the overall health risk.

Methods

A cross-sectional study was conducted among 2,306 school children in Guangzhou and Shenzhen. Questionnaire data on respiratory symptoms and diseases were collected along with sociodemographic and residential environmental information. A subset of children (N=987) were measured for their lung function. A random forest algorithm was applied to screen the top-ranked indoor environmental exposure variables and to form a composite index for cumulative risk of indoor pollution (CRIP). Logistic regressions were conducted to analyze the independent effect of single indoor environmental risk factors and the combined effect of CRIP on children’s respiratory health. Multiple linear regressions were used to examine the independent and combined effects of indoor environmental exposure on lung function.

Results

We found that home dampness and molds as well as environmental tobacco smoke (ETS) were significantly and independently associated with increased prevalence of children’s respiratory symptoms and diseases and with reduced lung function. A higher CRIP level was significantly associated with increased risk of cough with cold (OR =1.37, 95% CI: 1.05–1.79) and wheeze (OR =2.71, 95% CI: 1.16–6.34). A higher CRIP level was also associated with reduced lung function measured as FVC, FEV₁, PEF, FEF_25%, FEF_25–75% and VC.

Conclusions

In children living in the subtropical region of the Pearl River Delta, home dampness and the presence of mold as well as ETS were individual risk factors for children’s respiratory health. The composite CRIP index was associated with respiratory symptoms and lung function, suggesting the utility of this index for predicting the combined effects of multiple risk factors.

Precision Medicine for Neonatal Sepsis

Neonatal sepsis remains a significant cause of morbidity and mortality especially in the preterm infant population. The ability to promptly and accurately diagnose neonatal sepsis based on clinical evaluation and laboratory blood tests remains challenging. Advances in high-throughput molecular technologies have increased investigations into the utility of transcriptomic, proteomic and metabolomic approaches as diagnostic tools for neonatal sepsis. A systems-level understanding of neonatal sepsis, obtained by using omics-based technologies (at the transcriptome, proteome or metabolome level), may lead to new diagnostic tools for neonatal sepsis. In particular, recent omic-based studies have identified distinct transcriptional signatures and metabolic or proteomic biomarkers associated with sepsis. Despite the emerging need for a systems biology approach, future studies have to address the challenges of integrating multi-omic data with laboratory and clinical meta-data in order to translate outcomes into precision medicine for neonatal sepsis. Omics-based analytical approaches may advance diagnostic tools for neonatal sepsis. More research is needed to validate the recent systems biology findings in order to integrate multi-dimensional data (clinical, laboratory and multi-omic) for future translation into precision medicine for neonatal sepsis. This review will discuss the possible applications of omics-based analyses for identification of new biomarkers and diagnostic signatures for neonatal sepsis, focusing on the immune-compromised preterm infant and considerations for clinical translation.

An advanced fragment analysis-based individualized subtype classification of pediatric acute lymphoblastic leukemia

Pediatric acute lymphoblastic leukemia (ALL) is the most common neoplasm and one of the primary causes of death in children. Its treatment is highly dependent on the correct classification of subtype. Previously, we developed a microarray-based subtype classifier based on the relative expression levels of 62 marker genes, which can predict 7 different ALL subtypes with an accuracy as high as 97% in completely independent samples. Because the classifier is based on gene expression rank values rather than actual values, the classifier enables an individualized diagnosis, without the need to reference the background distribution of the marker genes in a large number of other samples, and also enables cross platform application. Here, we demonstrate that the classifier can be extended from a microarray-based technology to a multiplex qPCR-based technology using the same set of marker genes as the advanced fragment analysis (AFA). Compared to microarray assays, the new assay system makes the convenient, low cost and individualized subtype diagnosis of pediatric ALL a reality and is clinically applicable, particularly in developing countries.

Emerging technologies in paediatric leukaemia

Genetic changes, in particular chromosomal aberrations, are a hallmark of acute lymphoblastic lymphoma (ALL) and accurate detection of them is important in ensuring assignment to the appropriate drug protocol. Our ability to detect these genetic changes has been somewhat limited in the past due to the necessity to analyse mitotically active cells by conventional G-banded metaphase analysis and by mutational analysis of individual genes. Advances in technology include high resolution, microarray-based techniques that permit examination of the whole genome. Here we will review the current available methodology and discuss how the technology is being integrated into the diagnostic setting.

Genome analysis of Legionella pneumophila strains using a mixed-genome microarray

BACKGROUND

Legionella, the causative agent for Legionnaires' disease, is ubiquitous in both natural and man-made aquatic environments. The distribution of Legionella genotypes within clinical strains is significantly different from that found in environmental strains. Developing novel genotypic methods that offer the ability to distinguish clinical from environmental strains could help to focus on more relevant (virulent) Legionella species in control efforts. Mixed-genome microarray data can be used to perform a comparative-genome analysis of strain collections, and advanced statistical approaches, such as the Random Forest algorithm are available to process these data.

METHODS

Microarray analysis was performed on a collection of 222 Legionella pneumophila strains, which included patient-derived strains from notified cases in The Netherlands in the period 2002-2006 and the environmental strains that were collected during the source investigation for those patients within the Dutch National Legionella Outbreak Detection Programme. The Random Forest algorithm combined with a logistic regression model was used to select predictive markers and to construct a predictive model that could discriminate between strains from different origin: clinical or environmental.

RESULTS

Four genetic markers were selected that correctly predicted 96% of the clinical strains and 66% of the environmental strains collected within the Dutch National Legionella Outbreak Detection Programme.

CONCLUSIONS

The Random Forest algorithm is well suited for the development of prediction models that use mixed-genome microarray data to discriminate between Legionella strains from different origin. The identification of these predictive genetic markers could offer the possibility to identify virulence factors within the Legionella genome, which in the future may be implemented in the daily practice of controlling Legionella in the public health environment.

Epigenetic markers of prostate cancer in plasma circulating DNA

Epigenetic differences are a common feature of many diseases, including cancer, and disease-associated changes have even been detected in bodily fluids. DNA modification studies in circulating DNA (cirDNA) may lead to the development of specific non-invasive biomarkers. To test this hypothesis, we investigated cirDNA modifications in prostate cancer patients with locally confined disease (n = 19), in patients with benign prostate hyperplasias (n = 20) and in men without any known prostate disease (n = 20). This initial discovery screen identified 39 disease-associated changes in cirDNA modification, and seven of these were validated using the sodium bisulfite-based mapping of modified cytosines in both the discovery cohort and an independent 38-patient validation cohort. In particular, we showed that the DNA modification of regions adjacent to the gene encoding ring finger protein 219 distinguished prostate cancer from benign hyperplasias with good sensitivity (61%) and specificity (71%). We also showed that repetitive sequences detected in this study were meaningful, as they indicated a highly statistically significant loss of DNA at the pericentromeric region of chromosome 10 in prostate cancer patients (p = 1.8 × 10(-6)). Based on these strong univariate results, we applied machine-learning techniques to develop a multi-locus biomarker that correctly distinguished prostate cancer samples from unaffected controls with 72% accuracy. Lastly, we used systems biology techniques to integrate our data with publicly available DNA modification and transcriptomic data from primary prostate tumors, thereby prioritizing genes for further studies. These data suggest that cirDNA epigenomics are promising source for non-invasive biomarkers.

The projection score--an evaluation criterion for variable subset selection in PCA visualization

BACKGROUND

In many scientific domains, it is becoming increasingly common to collect high-dimensional data sets, often with an exploratory aim, to generate new and relevant hypotheses. The exploratory perspective often makes statistically guided visualization methods, such as Principal Component Analysis (PCA), the methods of choice. However, the clarity of the obtained visualizations, and thereby the potential to use them to formulate relevant hypotheses, may be confounded by the presence of the many non-informative variables. For microarray data, more easily interpretable visualizations are often obtained by filtering the variable set, for example by removing the variables with the smallest variances or by only including the variables most highly related to a specific response. The resulting visualization may depend heavily on the inclusion criterion, that is, effectively the number of retained variables. To our knowledge, there exists no objective method for determining the optimal inclusion criterion in the context of visualization.

RESULTS

We present the projection score, which is a straightforward, intuitively appealing measure of the informativeness of a variable subset with respect to PCA visualization. This measure can be universally applied to find suitable inclusion criteria for any type of variable filtering. We apply the presented measure to find optimal variable subsets for different filtering methods in both microarray data sets and synthetic data sets. We note also that the projection score can be applied in general contexts, to compare the informativeness of any variable subsets with respect to visualization by PCA.

CONCLUSIONS

We conclude that the projection score provides an easily interpretable and universally applicable measure of the informativeness of a variable subset with respect to visualization by PCA, that can be used to systematically find the most interpretable PCA visualization in practical exploratory analysis.

Gene expression signature of lymphocyte in acute lymphoblastic leukemia patients immediately after total body irradiation

Molecular mechanisms involved in TBI preconditioning before allogenetic transplantation remain unclear. To elucidate possible signaling pathway in it, gene expression profiles of peripheral lymphocytes were compared between samples 24h after each 4.5 Gy total body irradiation treatment (total 9 Gy) from 4 adult ALL patients. 478 significant expressed genes and three unique patterns were identified. Of these, a dominant progressively repressed expression of genes involved in ubiquitin-dependent process and repressed expression only at 9 Gy of genes involved in allograft rejection and graft-versus-host disease pathways were observed. The results suggest these pathways may play important roles for subsequent transplantation.

Gene-based outcome prediction in multiple cohorts of pediatric T-cell acute lymphoblastic leukemia: a Children's Oncology Group study

Background

Continuous complete clinical remission in T-cell acute lymphoblastic leukemia (T-ALL) is now approaching 80% due to the implementation of aggressive chemotherapy protocols but patients that relapse continue to have a poor prognosis. Such patients could benefit from augmented therapy if their clinical outcome could be more accurately predicted at the time of diagnosis. Gene expression profiling offers the potential to identify additional prognostic markers but has had limited success in generating robust signatures that predict outcome across multiple patient cohorts. This study aimed to identify robust gene classifiers that could be used for the accurate prediction of relapse in independent cohorts and across different experimental platforms.

Results

Using HG-U133Plus2 microarrays we modeled a five-gene classifier (5-GC) that accurately predicted clinical outcome in a cohort of 50 T-ALL patients. The 5-GC was further tested against three independent cohorts of T-ALL patients, using either qRT-PCR or microarray gene expression, and could predict patients with significantly adverse clinical outcome in each. The 5-GC featured the interleukin-7 receptor (IL-7R), low-expression of which was independently predictive of relapse in T-ALL patients. In T-ALL cell lines, low IL-7R expression was correlated with diminished growth response to IL-7 and enhanced glucocorticoid resistance. Analysis of biological pathways identified the NF-κB and Wnt pathways, and the cell adhesion receptor family (particularly integrins) as being predictive of relapse. Outcome modeling using genes from these pathways identified patients with significantly worse relapse-free survival in each T-ALL cohort.

Conclusions

We have used two different approaches to identify, for the first time, robust gene signatures that can successfully discriminate relapse and CCR patients at the time of diagnosis across multiple patient cohorts and platforms. Such genes and pathways represent markers for improved patient risk stratification and potential targets for novel T-ALL therapies.

Validation of a mouse xenograft model system for gene expression analysis of human acute lymphoblastic leukaemia

BACKGROUND

Pre-clinical models that effectively recapitulate human disease are critical for expanding our knowledge of cancer biology and drug resistance mechanisms. For haematological malignancies, the non-obese diabetic/severe combined immunodeficient (NOD/SCID) mouse is one of the most successful models to study paediatric acute lymphoblastic leukaemia (ALL). However, for this model to be effective for studying engraftment and therapy responses at the whole genome level, careful molecular characterisation is essential.

RESULTS

Here, we sought to validate species-specific gene expression profiling in the high engraftment continuous ALL NOD/SCID xenograft. Using the human Affymetrix whole transcript platform we analysed transcriptional profiles from engrafted tissues without prior cell separation of mouse cells and found it to return highly reproducible profiles in xenografts from individual mice. The model was further tested with experimental mixtures of human and mouse cells, demonstrating that the presence of mouse cells does not significantly skew expression profiles when xenografts contain 90% or more human cells. In addition, we present a novel in silico and experimental masking approach to identify probes and transcript clusters susceptible to cross-species hybridisation.

CONCLUSIONS

We demonstrate species-specific transcriptional profiles can be obtained from xenografts when high levels of engraftment are achieved or with the application of transcript cluster masks. Importantly, this masking approach can be applied and adapted to other xenograft models where human tissue infiltration is lower. This model provides a powerful platform for identifying genes and pathways associated with ALL disease progression and response to therapy in vivo.

Genome-wide expression analysis of paired diagnosis-relapse samples in ALL indicates involvement of pathways related to DNA replication, cell cycle and DNA repair, independent of immune phenotype

Almost a quarter of pediatric patients with acute lymphoblastic leukemia (ALL) suffer from relapses. The biological mechanisms underlying therapy response and development of relapses have remained unclear. In an attempt to better understand this phenomenon, we have analyzed 41 matched diagnosis-relapse pairs of ALL patients using genome-wide expression arrays (82 arrays) on purified leukemic cells. In roughly half of the patients, very few differences between diagnosis and relapse samples were found ('stable group'), suggesting that mostly extra-leukemic factors (for example, drug distribution, drug metabolism, compliance) contributed to the relapse. Therefore, we focused our further analysis on 20 sample pairs with clear differences in gene expression ('skewed group'), reasoning that these would allow us to better study the biological mechanisms underlying relapsed ALL. After finding the differences between diagnosis and relapse pairs in this group, we identified four major gene clusters corresponding to several pathways associated with changes in cell cycle, DNA replication, recombination and repair, as well as B-cell developmental genes. We also identified cancer genes commonly associated with colon carcinomas and ubiquitination to be upregulated in relapsed ALL. Thus, about half of the relapses are due to the selection or emergence of a clone with deregulated expression of genes involved in pathways that regulate B-cell signaling, development, cell cycle, cellular division and replication.

Early T cell differentiation lessons from T-cell acute lymphoblastic leukemia

T cells develop from bone marrow-derived self-renewing hematopoietic stem cells (HSC). Upon entering the thymus, these cells undergo progressive commitment and differentiation driven by the thymic stroma and the pre-T cell receptor (pre-TCR). These processes are disrupted in T-cell acute lymphoblastic leukemia (T-ALL). More than 70% of recurring chromosomal rearrangements in T-ALL activate the expression of oncogenic transcription factors, belonging mostly to three families, basic helix-loop-helix (bHLH), homeobox (HOX), and c-MYB. This prevalence is indicative of their importance in the T lineage, and their dominant mechanisms of transformation. For example, bHLH oncoproteins inhibit E2A and HEB, revealing their tumor suppressor function in the thymus. The induction of T-ALL, nonetheless, requires collaboration with constitutive NOTCH1 signaling and the pre-TCR, as well as loss-of-function mutations for CDKN2A and PTEN. Significantly, NOTCH1, the pre-TCR pathway, and E2A/HEB proteins control critical checkpoints and branchpoints in early thymocyte development whereas several oncogenic transcription factors, HOXA9, c-MYB, SCL, and LYL-1 control HSC self-renewal. Together, these genetic lesions alter key regulatory processes in the cell, favoring self-renewal and subvert the normal control of thymocyte homeostasis.

Gene expression–based classification and regulatory networks of pediatric acute lymphoblastic leukemia

Pediatric acute lymphoblastic leukemia (ALL) contains cytogenetically distinct subtypes that respond differently to cytotoxic drugs. Subtype classification can be also achieved through gene expression profiling. However, how to apply such classifiers to a single patient and correctly diagnose the disease subtype in an independent patient group has not been addressed. Furthermore, the underlying regulatory mechanisms responsible for the subtype-specific gene expression patterns are still largely unknown. Here, by combining 3 published microarray datasets on 535 mostly white children's samples and generating a new dataset on 100 Chinese children's ALL samples, we were able to (1) identify a 62-gene classifier with 97.6% accuracy from the white children's samples and validated it on the completely independent set of 100 Chinese samples, and (2) uncover potential regulatory networks of ALL subtypes. The classifier we identified was, thus far, the only one that could be applied directly to a single sample and that sustained validation in a large independent patient group. Our results also suggest that the etiology of ALL is largely the same among different ethnic groups, and that the transcription factor hubs in the predicted regulatory network might play important roles in regulating gene expression and development of ALL.

Predicting sulfotyrosine sites using the random forest algorithm with significantly improved prediction accuracy

BACKGROUND

Tyrosine sulfation is one of the most important posttranslational modifications. Due to its relevance to various disease developments, tyrosine sulfation has become the target for drug design. In order to facilitate efficient drug design, accurate prediction of sulfotyrosine sites is desirable. A predictor published seven years ago has been very successful with claimed prediction accuracy of 98%. However, it has a particularly low sensitivity when predicting sulfotyrosine sites in some newly sequenced proteins.

RESULTS

A new approach has been developed for predicting sulfotyrosine sites using the random forest algorithm after a careful evaluation of seven machine learning algorithms. Peptides are formed by consecutive residues symmetrically flanking tyrosine sites. They are then encoded using an amino acid hydrophobicity scale. This new approach has increased the sensitivity by 22%, the specificity by 3%, and the total prediction accuracy by 10% compared with the previous predictor using the same blind data. Meanwhile, both negative and positive predictive powers have been increased by 9%. In addition, the random forest model has an excellent feature for ranking the residues flanking tyrosine sites, hence providing more information for further investigating the tyrosine sulfation mechanism. A web tool has been implemented at http://ecsb.ex.ac.uk/sulfotyrosine for public use.

CONCLUSION

The random forest algorithm is able to deliver a better model compared with the Hidden Markov Model, the support vector machine, artificial neural networks, and others for predicting sulfotyrosine sites. The success shows that the random forest algorithm together with an amino acid hydrophobicity scale encoding can be a good candidate for peptide classification.

Blood-based transcriptomics: leukemias and beyond

In 1999, Golub et al. proposed for the first time microarray-based transcriptional profiling to be used as a new technology for the differential diagnosis of acute myeloid leukemias and acute lymphocytic leukemias. This very preliminary study sparked great enthusiasm beyond the leukemias. Over the last 10 years, numerous studies addressed the use of gene expression profiling of peripheral blood from patients with malignancies, infectious diseases, autoimmunity and even cardiovascular diseases. Despite this great effort, no single test has yet been established using microarray-based transcriptional profiling of peripheral blood. Here we highlight the advances in the field of blood transcriptomics during the last 10 years and also critically discuss the issues that need to be resolved before blood transcriptomics will become part of daily diagnostics in the leukemias, as well as in other diseases showing involvement of peripheral blood.

Prediction of relapse in paediatric pre-B acute lymphoblastic leukaemia using a three-gene risk index

Despite high cure rates 25% of children with acute lymphoblastic leukaemia (ALL) relapse and have dismal outcome. Crucially, many are currently stratified as standard risk (SR) and additional markers to improve patient stratification are required. Here we have used diagnostic bone marrow specimens from 101 children with pre-B ALL to examine the use of gene expression profiles (GEP) as predictors of long-term clinical outcome. Patients were divided into two cohorts for model development and validation based on availability of specimen material. Initially, GEP from 55 patients with sufficient material were analysed using HG-U133A microarrays, identifying an 18-gene classifier (GC) that was more predictive of outcome than conventional prognostic parameters. After feature selection and validation of expression levels by quantitative reverse transcription polymerase chain reaction (qRT-PCR), a three-gene qRT-PCR risk index [glutamine synthetase (GLUL), ornithine decarboxylase antizyme inhibitor (AZIN), immunoglobulin J chain (IGJ)] was developed that predicted outcome with an accuracy of 89% in the array cohort and 87% in the independent validation cohort. The data demonstrate the feasibility of using GEP to improve risk stratification in childhood ALL. This is particularly important for the identification of patients destined to relapse despite their current stratification as SR, as more intensive front-line treatment options for these individuals are already available.

SDED: a novel filter method for cancer-related gene selection

Gene selection is to detect the most significantly expressed genes under different conditions expression data. The current challenge in gene selection is the comparison of a large number of genes with limited patient samples. Thus it is trivial task in simple statistical analysis. Various statistical measurements are adopted by filter methods applied in gene selection studies. Their ability to discriminate phenotypes is crucial in classification and selection. Here we describe the standard deviation error distribution (SDED) method for gene selection. It utilizes variations within-class and among-class in gene expression data. We tested the method using 4 leukemia datasets available in the public domain. The method was compared with the GS2 and CHO methods. The Prediction accuracies by SDED are better than both GS2 and CHO for different datasets. These are 0.8-4.2% and 1.6-8.4% more that in GS2 and CHO. The related OMIM annotations and KEGG pathways analyses verified that SDED can pick out more 4.0% and 6.1% genes with biological significance than GS2 and CHO, respectively.

High expression of connective tissue growth factor in pre-B acute lymphoblastic leukaemia

In recent years microarrays have been used extensively to characterize gene expression in acute lymphoblastic leukaemia (ALL). Few studies, however, have analysed normal haematopoietic cell populations to identify altered gene expression in ALL. We used oligonucleotide microarrays to compare the gene expression profile of paediatric precursor-B (pre-B) ALL specimens with two control cell populations, normal CD34(+) and CD19(+)IgM(-) cells, to focus on genes linked to leukemogenesis. A set of eight genes was identified with a ninefold higher average expression in ALL specimens compared with control cells. All of these genes were significantly deregulated in an independent cohort of 101 ALL specimens. One gene, connective tissue growth factor (CTGF, also known as CCN2), had exceptionally high expression, which was confirmed in three independent leukaemia studies. Further analysis of CTGF expression in ALL revealed exclusive expression in B-lineage, not T-lineage, ALL. Within B-lineage ALL approximately 75% of specimens were consistently positive for CTGF expression, however, specimens containing the E2A-PBX1 translocation showed low or no expression. Protein studies using Western blot analysis demonstrated the presence of CTGF in ALL cell-conditioned media. These findings indicate that CTGF is secreted by pre-B ALL cells and may play a role in the pathophysiology of this disease.

Identification of novel molecular prognostic markers for paediatric T-cell acute lymphoblastic leukaemia

In the last four decades the survival of patients with newly diagnosed childhood T-cell acute lymphoblastic leukaemia (T-ALL) has improved dramatically. In sharp contrast, relapsed T-ALL continues to confer a dismal prognosis. We sought to determine if gene expression profiling could uncover a signature of outcome for children with T-ALL. Using 12 patient specimens obtained before therapy started, we examined the gene expression profile by oligonucleotide microarrays. We identified three genes, CFLAR, NOTCH2 and BTG3, whose expression at the time of diagnosis accurately distinguished the patients according to disease outcome. These genes are involved in the regulation of apoptosis and cellular proliferation. The prognostic value of the three predictive genes was assessed in an independent cohort of 25 paediatric T-ALL patients using quantitative real-time reverse transcription polymerase chain reaction. Patients assigned to the adverse outcome group had a significantly higher cumulative incidence of relapse compared with patients assigned to the favourable outcome group (46% vs. 8%, P = 0.029). Five-year overall survival was also significantly worse in the patients assigned to the adverse outcome group (P = 0.0039). The independent influence of the 3-gene predictor was confirmed by multivariate analysis. Our study provides proof of principle that genome-wide expression profiling can detect novel molecular prognostic markers in paediatric T-ALL.