Classification of human tumors using gene expression profiles obtained after microarray analysis of fine-needle aspiration biopsy samples.

BACKGROUND: Gene expression profiling using gene-discovery, high-density microarray technologies is a powerful tool. One potential application is the development of tumor classifiers that predict the site of origin. For this technology to be relevant, however, it must be applicable to tumor biopsy samples, which most often are fine-needle aspiration biopsy (FNAB) samples.
METHODS: Surgically resected tumors were sampled by FNAB using different gauge needles. A portion of the excised tumor was also collected. RNA samples were extracted using standard techniques and the quality and quantity of the RNA samples were measured for each sample. Thirteen representative FNAB samples and two representative tissue samples were submitted for microarray analysis and then subjected to a tumor classifier.
RESULTS: Fourteen of 18 samples analyzed for quantity and quality of RNA yielded an adequate amount of RNA (> 1 microg total RNA). Tumor type contributed to the RNA yield because one of the four inadequate samples was retrieved from a patient with lobular carcinoma of the breast and the other three samples were retrieved from patients with retroperitoneal sarcomas. Of the 13 samples submitted for microarray analysis, 9 were classified correctly as to tumor type using a tissue-based tumor classifier.
CONCLUSIONS: The authors demonstrated that FNABs reproducibly obtained an adequate amount of RNA for microarray analysis when a standardized collection procedure was used. Furthermore, the samples generated interpretable gene expression profiles that could be matched accurately with a tumor classifier established on tissue specimens. The current study showed that FNAB produced adequate material for microarray analysis when utilizing a standardized collection procedure. 2005 American Cancer Society.