Statistical framework for gene expression data analysis.

DNA (mRNA) microarray, a highly promising technique with a variety of applications, can yield a wealth of data about each sample, well beyond the reach of every individual's comprehension. A need exists for statistical approaches that reliably eliminate insufficient and uninformative genes (probe sets) from further analysis while keeping all essentially important genes. This procedure does call for in-depth knowledge of the biological system to analyze. We conduct a comparative study of several statistical approaches on our own breast cancer Affymetrix microarray datasets. The strategy is designed primarily as a filter to select subsets of genes relevant for classification. We outline a general framework based on different statistical algorithms for determining a high-performing multigene predictor of response to the preoperative treatment of patients. We hope that our approach will provide straightforward and useful practical guidance for identification of genes, which can discriminate between biologically relevant classes in microarray datasets.

Gene expression in acute Stanford type A dissection: a comparative microarray study.

BACKGROUND: We compared gene expression profiles in acutely dissected aorta with those in normal control aorta. MATERIALS AND METHODS: Ascending aorta specimen from patients with an acute Stanford A-dissection were taken during surgery and compared with those from normal ascending aorta from multiorgan donors using the BD Atlas Human1.2 Array I, BD Atlas Human Cardiovascular Array and the Affymetrix HG-U133A GeneChip. For analysis only genes with strong signals of more than 70 percent of the mean signal of all spots on the array were accepted as being expressed. Quantitative real-time polymerase chain reaction (RT-PCR) was used to confirm regulation of expression of a subset of 24 genes known to be involved in aortic structure and function. RESULTS: According to our definition expression profiling of aorta tissue specimens revealed an expression of 19.1% to 23.5% of the genes listed on the arrays. Of those 15.7% to 28.9% were differently expressed in dissected and control aorta specimens. Several genes that encode for extracellular matrix components such as collagen IV alpha2 and -alpha5, collagen VI alpha3, collagen XIV alpha1, collagen XVIII alpha1 and elastin were down-regulated in aortic dissection, whereas levels of matrix metalloproteinases-11, -14 and -19 were increased. Some genes coding for cell to cell adhesion, cell to matrix signaling (e.g., polycystin1 and -2), cytoskeleton, as well as several myofibrillar genes (e.g., alpha-actinin, tropomyosin, gelsolin) were found to be down-regulated. Not surprisingly, some genes associated with chronic inflammation such as interleukin -2, -6 and -8, were up-regulated in dissection. CONCLUSION: Our results demonstrate the complexity of the dissecting process on a molecular level. Genes coding for the integrity and strength of the aortic wall were down-regulated whereas components of inflammatory response were up-regulated. Altered patterns of gene expression indicate a pre-existing structural failure, which is probably a consequence of insufficient remodeling of the aortic wall resulting in further aortic dissection.

Predictors of primary breast cancers responsiveness to preoperative epirubicin/cyclophosphamide-based chemotherapy: translation of microarray data into clinically useful predictive signatures.

BACKGROUND: Our goal was to identify gene signatures predictive of response to preoperative systemic chemotherapy (PST) with epirubicin/cyclophosphamide (EC) in patients with primary breast cancer. METHODS: Needle biopsies were obtained pre-treatment from 83 patients with breast cancer and mRNA was profiled on Affymetrix HG-U133A arrays. Response ranged from pathologically confirmed complete remission (pCR), to partial remission (PR), to stable or progressive disease, "No Change" (NC). A primary analysis was performed in breast tissue samples from 56 patients and 5 normal healthy individuals as a training cohort for predictive marker identification. Gene signatures identifying individuals most likely to respond completely to PST-EC were extracted by combining several statistical methods and filtering criteria. In order to optimize prediction of non responding tumors Student's t-test and Wilcoxon test were also applied. An independent cohort of 27 patients was used to challenge the predictive signatures. A k-Nearest neighbor algorithm as well as two independent linear partial least squares determinant analysis (PLS-DA) models based on the training cohort were selected for classification of the test samples. The average specificity of these predictions was greater than 74% for pCR, 100% for PR and greater than 62% for NC. All three classification models could identify all pCR cases. RESULTS: The differential expression of 59 genes in the training and the test cohort demonstrated capability to predict response to PST-EC treatment. Based on the training cohort a classifier was constructed following a decision tree. First, a transcriptional profile capable to distinguish cancerous from normal tissue was identified. Then, a "favorable outcome signature" (31 genes) and a "poor outcome signature" (26 genes) were extracted from the cancer specific signatures. This stepwise implementation could predict pCR and distinguish between NC and PR in a subsequent set of patients. Both PLS-DA models were implemented to discriminate all three response classes in one step. CONCLUSION: In this study signatures were identified capable to predict clinical outcome in an independent set of primary breast cancer patients undergoing PST-EC.

Predictive biological markers for response of invasive breast cancer to anthracycline/cyclophosphamide-based primary (radio-)chemotherapy.

BACKGROUND: The role of biological markers for the prediction of neoadjuvant chemotherapy and radio-chemotherapy may be evaluated using pathological complete response [pCR] in patients with invasive breast cancer. MATERIALS AND METHODS: To investigate this, pre-treatment biopsies from 517 patients with locally advanced breast cancer were analyzed for expression of estrogen receptor [ER], progesterone receptor [PgR], Her-2/neu, epidermal growth factor receptor [EGF-R], p53, Bcl-2 and MIB-1 by immunohistochemistry [IHC], and these data were compared to the pathological response after preoperative epirubicine/cyclophosphamide [EC] chemotherapy (+/- radiotherapy). RESULTS: pCR was more frequent (28.30%, 56/198) in tumors that received radio-chemotherapy compared to chemotherapy alone (11.9%, 38/319, p < 0.0001). Patients with high grading, lower ER, PgR, Bcl-2 or a higher proliferation had a significantly greater benefit from chemotherapy. The overexpressions of Her2/neu or EGF-R were weakly correlated to pCR, while p53 staining did not have any predictive value. Younger patients, with negative PgR and high proliferation index, had the highest benefit from EC therapy (56% pCR). The different multivariate indices of binary regression, PLS-DA and SIMCA, had similar predictive quality and were slightly superior to univariate factors. CONCLUSION: This study emphasizes the value of traditional biological markers and Bcl-2 for use in the individual selection of a primary therapy regimen.

Immediate gene expression changes after the first course of neoadjuvant chemotherapy in patients with primary breast cancer disease.

PURPOSE: Our goal was to identify genes undergoing expressional changes shortly after the beginning of neoadjuvant chemotherapy for primary breast cancer. EXPERIMENTAL DESIGN: The biopsies were taken from patients with primary breast cancer prior to any treatment and 24 hours after the beginning of the neoadjuvant chemotherapy. Expression analyses from matched pair samples representing 25 patients were carried out with Clontech filter arrays. A subcohort of those 25 paired samples were additionally analyzed with the Affymetrix GeneChip platform. All of the transcripts from both platforms were queried for expressional changes. RESULTS: Performing hierarchical cluster analysis, we clustered pre- and posttreatment samples from individual patients more closely to each other than the samples taken from different patients. This reflects the rather low number of transcripts responding directly to the drugs used. Although transcriptional drug response occurring during therapy differed between individual patients, two genes (p21(WAF1/CIP1) and MIC-1) were up-regulated in posttreatment samples. This could be validated by semiquantitative and real-time reverse transcription-PCR. Partial least- discriminant analysis based on approximately 25 genes independently identified by either Clontech or Affymetrix platforms could clearly discriminate pre- and posttreatment samples. However, correlation of certain gene expression levels as well as of differential patterns and clusters as determined by a different platform was not always satisfying. CONCLUSIONS: This study has demonstrated the potential of monitoring posttreatment changes in gene expression as a measure of the pharmacodynamics of drugs. As a clinical laboratory model, it can be useful to identify patients with sensitive and reactive tumors and to help for optimized choice for sequential therapy and obviously improve relapse- free and overall survival.

Identifying superficial, muscle-invasive, and metastasizing transitional cell carcinoma of the bladder: use of cDNA array analysis of gene expression profiles.

PURPOSE: Expression profiling by DNA microarray technology permits the identification of genes underlying clinical heterogeneity of bladder cancer and which might contribute to disease progression, thereby improving assessment of treatment and prediction of patient outcome. EXPERIMENTAL DESIGN: Invasive (20) and superficial (22) human bladder tumors from 34 patients with known outcome regarding disease recurrence and progression were analyzed by filter-based cDNA arrays (Atlas Human Cancer 1.2; BD Biosciences Clontech) containing 1185 genes. For 9 genes, array data were confirmed using real-time reverse transcription-PCR. Additionally, Atlas array data were validated using Affymetrix GeneChip oligonucleotide arrays with 22,283 human gene fragments and expressed sequence tags sequences in a subset of three superficial and six invasive bladder tumors. RESULTS: A two-way clustering algorithm using different subsets of gene expression data, including a subset of 41 genes validated by the oligonucleotide array (Affymetrix), classified tumor samples according to clinical outcome as superficial, invasive, or metastasizing. Furthermore, (a) a clonal origin of superficial tumors, (b) highly similar gene expression patterns in different areas of invasive tumors, and (c) an invasive-like pattern was observed in bladder mucosas derived from patients with locally advanced disease. Several gene clusters that characterized invasive or superficial tumors were identified. In superficial bladder tumors, increased mRNA levels of genes encoding transcription factors, molecules involved in protein synthesis and metabolism, and some proteins involved into cell cycle progression and differentiation were observed, whereas transcripts for immune, extracellular matrix, adhesion, peritumoral stroma and muscle tissue components, proliferation, and cell cycle controllers were up-regulated in invasive tumors. CONCLUSIONS: Gene expression profiling of human bladder cancers provides insight into the biology of bladder cancer progression and identifies patients with distinct clinical phenotypes.

Gene expression profiling of advanced head and neck squamous cell carcinomas and two squamous cell carcinoma cell lines under radio/chemotherapy using cDNA arrays.

BACKGROUND AND PURPOSE: The response of squamous cell carcinomas of the head and neck (HNSCC) to radio/chemotherapy is accompanied by complex changes in patterns of gene expression. It is highly probable that a better understanding of molecular and genetic changes can help to optimize the treatment of HNSCC. cDNA arrays provide a powerful tool for high-throughput monitoring of gene expression in small clinical specimens. MATERIALS AND METHODS: We used tumour biopsies from four patients with HNSCC which have been taken prior to and during radio/chemotherapy. The patterns of gene expression obtained from clinical samples were compared with gene expression profiles of two squamous cell carcinoma cell lines (FaDU and UD-7A). RESULTS: The experimental data analysis revealed changes in expression levels of several genes during radio/chemotherapy. Despite treatment, independent samples taken from the same cell line or tumour in situ were more similar to each other than either was to other specimens. The data indicate a high gene heterogeneity of HNSCC that is preserved during treatment. CONCLUSIONS: From our preliminary results we conclude that the cDNA array experimental approach can detect differences in gene expression between treated and untreated small tumour biopsies, as well as inter-individual differences in expression profiles between HNSCC tumours. The examination of a greater sample size will be needed to make this preliminary evaluation useful to elucidate the functional significance of individual genes which exhibit altered levels of expression under radiation therapy.

Breast cancer expression profiling: the impact of microarray testing on clinical decision making.

The available clinical prognostic tools show an obvious limitation in predicting the outcome of breast cancer patients, and pathological features cannot classify tumours accurately. Microarray-based molecular classification of breast tumours or selection of gene expression panels to improve risk prediction or treatment outcomes are thought to be theoretically superior to established clinical and pathological criteria, based on guidelines such as the St Gallen and National Institute of Health consensus, or which use specific prognostic tools, such as the Nottingham Prognostic Index or Adjuvant-Online algorithm. Although two diagnostic tests based on gene expression profiling of breast cancer are commercially available, a new molecular classification and molecular forecasting of breast cancer based on expression profiling cannot outperform the standard tumour diagnostic at present. This review focuses on some important problems in the practical application of molecular profiling of breast cancer for clinical purposes.

Generalization of DNA microarray dispersion properties: microarray equivalent of t-distribution.

BACKGROUND: DNA microarrays are a powerful technology that can provide a wealth of gene expression data for disease studies, drug development, and a wide scope of other investigations. Because of the large volume and inherent variability of DNA microarray data, many new statistical methods have been developed for evaluating the significance of the observed differences in gene expression. However, until now little attention has been given to the characterization of dispersion of DNA microarray data. RESULTS: Here we examine the expression data obtained from 682 Affymetrix GeneChips with 22 different types and we demonstrate that the Gaussian (normal) frequency distribution is characteristic for the variability of gene expression values. However, typically 5 to 15% of the samples deviate from normality. Furthermore, it is shown that the frequency distributions of the difference of expression in subsets of ordered, consecutive pairs of genes (consecutive samples) in pair-wise comparisons of replicate experiments are also normal. We describe a consecutive sampling method, which is employed to calculate the characteristic function approximating standard deviation and show that the standard deviation derived from the consecutive samples is equivalent to the standard deviation obtained from individual genes. Finally, we determine the boundaries of probability intervals and demonstrate that the coefficients defining the intervals are independent of sample characteristics, variability of data, laboratory conditions and type of chips. These coefficients are very closely correlated with Student's t-distribution. CONCLUSION: In this study we ascertained that the non-systematic variations possess Gaussian distribution, determined the probability intervals and demonstrated that the K(alpha) coefficients defining these intervals are invariant; these coefficients offer a convenient universal measure of dispersion of data. The fact that the K(alpha) distributions are so close to t-distribution and independent of conditions and type of arrays suggests that the quantitative data provided by Affymetrix technology give "true" representation of physical processes, involved in measurement of RNA abundance. REVIEWERS: This article was reviewed by Yoav Gilad (nominated by Doron Lancet), Sach Mukherjee (nominated by Sandrine Dudoit) and Amir Niknejad and Shmuel Friedland (nominated by Neil Smalheiser).

Gene expression profiling identifies significant differences between the molecular phenotypes of bone marrow-derived and circulating human CD34+ hematopoietic stem cells.

CD34+ hematopoietic stem cells are used clinically to support cytotoxic therapy, and recent studies raised hope that they could even serve as a cellular source for nonhematopoietic tissue engineering. Here, we examined in 18 volunteers the gene expressions of 1185 genes in highly enriched bone marrow CD34+ (BM-CD34+) or granulocyte-colony-stimulating factor-mobilized peripheral blood CD34+ (PB-CD34+) cells by means of cDNA array technology to identify molecular causes underlying the functional differences between circulating and sedentary hematopoietic stem and progenitor cells. In total, 65 genes were significantly differentially expressed. Greater cell cycle and DNA synthesis activity of BM-CD34+ than PB-CD34+ cells were reflected by the 2- to 5-fold higher expression of 9 genes involved in cell cycle progression, 11 genes regulating DNA synthesis, and cell cycle-initiating transcription factor E2F-1. Conversely, 9 other transcription factors, including the differentiation blocking GATA2 and N-myc, were expressed 2 to 3 times higher in PB-CD34+ cells than in BM-CD34+ cells. Expression of 5 apoptosis driving genes was also 2 to 3 times greater in PB-CD34+ cells, reflecting a higher apoptotic activity. In summary, our study provides a gene expression profile of primary human CD34+ hematopoietic cells of the blood and marrow. Our data molecularly confirm and explain the finding that CD34+ cells residing in the bone marrow cycle more rapidly, whereas circulating CD34+ cells consist of a higher number of quiescent stem and progenitor cells. Moreover, our data provide novel molecular insight into stem cell physiology.