Luca Del Giacco | Search PubMed

PubMed2004

Development of a new reference standard for microarray experiments.

Often microarray studies require a reference to indirectly compare the samples under observation. References based on pooled RNA from different cell lines have already been described (here referred to as RNA-R), but they usually do not exhaustively represent the set of genes printed on a chip, thus requiring many adjustments during the analyses. A reference could also be generated in vitro transcribing the collection of cDNA clones printed on the microarray in use (here referred to as T3-R). Here we describe an alternative and simpler PCR-based methodology to construct a similar reference (Chip-R), and we extensively test and compare it to both RNA-R and T3-R. The use of both Chip-R and T3-R dramatically increases the number of signals on the slides and gives more reproducible results than RNA-R. Each reference preparation is also evaluated in a simple microarray experiment comparing two different RNA populations. Our results show that the introduction of a reference always interferes with the analysis. Indeed, the direct comparison is able to identify more up- or down-regulated genes than any reference-mediated analysis. However, if a reference has to be used, Chip-R and T3-R are able to guarantee more reliable results than RNA-R.

Calibration↗

PubMedDate not supplied

Microarrays in cancer research.

Microarray technology has presented the scientific community with a compelling approach that allows for simultaneous evaluation of all cellular processes at once. Cancer, being one of the most challenging diseases due to its polygenic nature, presents itself as a perfect candidate for evaluation by this approach. Several recent articles have provided significant insight into the strengths and limitations of microarrays. Nevertheless, there are strong indications that this approach will provide new molecular markers that could be used in diagnosis and prognosis of cancers. To achieve these goals it is essential that there is a seamless integration of clinical and molecular biological data that allows us to elucidate genes and pathways involved in various cancers. To this effect we are currently evaluating gene expression profiles in human brain, ovarian, breast and hematopoetic, lung, colorectal, head and neck and biliary tract cancers. To address the issues we have a joint team of scientists, doctors and computer scientists from two Virginia Universities and a major healthcare provider. The study has been divided into several focus groups that include; Tissue Bank Clinical & Pathology Laboratory Data, Chip Fabrication, QA/QC, Tissue Devitalization, Database Design and Data Analysis, using multiple microarray platforms. Currently over 300 consenting patients have been enrolled in the study with the largest number being that of breast cancer patients. Clinical data on each patient is being compiled into a secure and interactive relational database and integration of these data elements will be accomplished by a common programming interface. This clinical database contains several key parameters on each patient including demographic (risk factors, nutrition, co-morbidity, familial history), histopathology (non genetic predictors), tumor, treatment and follow-up information. Gene expression data derived from the tissue samples will be linked to this database, which allows us to query the data at multiple levels. The challenge of tissue acquisition and processing is of paramount importance to the success of this venture. A tissue devitalization timeline protocol was devised to ensure sample and RNA integrity. Stringent protocols are being employed to ascertain accurate tumor homogeneity, by serial dissection of each tumor sample at 10 microM frozen sections followed by histopathological evaluation. The multiple platforms being utilized in this study include Affimetrix, Oligo-Chips and custom-designed cDNA arrays. Selected RNA samples will be evaluated on each platform between the groups. Analysis steps will involve normalization and standardization of gene expression data followed by hierarchical clustering to determine co-regulation profiles. The aim of this conjoint effort is to elucidate pathways and genes involved in various cancers, resistance mechanisms, molecular markers for diagnosis and prognosis.

Data Collection↗