Misconceptions, challenges, uncertainty, and progress in guideline recommendations.

Although the quality of clinical practice guidelines has improved over the last decade, current guideline systems have limitations that reduce their validity and limit their acceptance. The Grading of Recommendations, Assessment, Development and Evaluation (GRADE) working group, a worldwide collaboration of guideline developers, methodologists, and clinicians, has constructed a system for developing guidelines that addresses these shortcomings. The system includes a transparent and rigorous methodology for rating the quality of evidence, an explicit balancing of benefits and harms of healthcare interventions, an explicit acknowledgement of the values and preferences that underlie the recommendations, and whether the intervention represents a wise use of resources. These four elements determine whether a recommendation is strong or weak. A guideline panel offers strong recommendations when virtually all informed patients would choose the same management strategy. Weak recommendations imply that choices will differ across the range of patient values and preferences. The GRADE system has been tested in multiple practice settings and for a large spectrum of questions, refined and re-evaluated to ensure that it captures the complex issues involved in evidence assessment and grading recommendations while maintaining simplicity and practicality. Many guideline organizations and medical societies have endorsed the system and adopted it for their guideline processes.

Gene expression profiling of human cancers.

DNA microarrays allow us to visualize simultaneously the expression of potentially all genes within a cell population or tissue sample-revealing the "transcriptome." The analysis of this type of data is commonly called "gene expression profiling" (GEP) because it provides a comprehensive picture of the pattern of gene expression in a particular biological sample. For this reason microarrays are revolutionizing life sciences research and are leading to the development of novel and powerful methods for investigating cancer biology, classifying cancers, and predicting clinical outcome of cancers. Several recent high-profile reports have revealed how clustering of GEP data can clearly identify clinically (and prognostically) important subtypes of cancer among patients considered by established clinicopathological criteria to have similar tumors. Accurate "prognostic signatures" can be obtained from GEP data, which represent relatively small numbers of genes. These signatures can be valuable in directing appropriate treatment and in predicting clinical outcome, and they generally outperform other systems based on clinical and histological criteria. In this paper the basic principles of DNA microarray technology and the different types of microarray platforms available will be introduced, and the power of the technique will be illustrated by reviewing some recent GEP studies on selected cancers, including a preliminary analysis of hepatocellular carcinoma from our Palermo laboratory. GEP is likely to be adopted in the future as a key decision-making tool in the clinical arena. However, several issues relating to data analysis, reproducibility, cross-comparability, validation, and cost need to be resolved before the technology can be adopted broadly in this context.