Pathological features of breast tissue samples do not allow oncologists to diagnose breast cancer with a sufficient degree of reliability, and determining the extent to which cancer cells have spread to the lymph nodes cannot enable them to make a sufficiently accurate prognosis. Microarrays could offer more precise diagnoses and prognoses, but their utility remains questionable, reported Olga Modlich and colleagues at the University of Düsseldorf, Germany, in a review of breast cancer gene profiling assays. Clinical laboratories, rather than companies, developed the gene profiling assays currently used in clinical environments because the Clinical Laboratory Improvement Act allowed them to do so without FDA approval. Two of those tests, a 21-gene panel and a 70-gene panel, have been licensed to Genomic Health Inc. of Redwood City, Calif., and Agendia of Amsterdam, the Netherlands, respectively. Every patient and tumor is different, so developing accurate microarrays remains challenging. Gene expression levels vary between patients and among tissues. Patients have inherently different relapse and survival rates, and varying responses to chemotherapy. Still, microarrays may be able to show initial responses to drugs and to predict whether a patient will develop drug resistance — factors that cannot be determined with other tests. The authors also considered microarrays’ quality and analysis. Self-made arrays tend to be of low quality, so the authors recommend using commercial ones instead. Microarrays often produce voluminous data that must be statistically analyzed, but the cluster analysis statistical method has led to a high number of false-positives, and supervised analysis has reflected irregularities in patient group composition. Ultimately, the reviewers doubt that gene profiling assays will be routinely used for breast cancer detection in the near future — at least until better statistical methods are developed and more clinical trials run. (Expert Opinion on Pharmacotherapy, October 2006, pp. 2069-2078.)