Recursive partitioning for tumor classification with gene expression microarray data

Precise classification of tumors is critically important for cancer diagnosis and treatment. It is also a scientifically challenging task. Recently, efforts have been made to use gene expression profiles to improve the precision of classification, with limited success. Using a published data set for purposes of comparison, we introduce a methodology based on classification trees and demonstrate that it is significantly more accurate for discriminating among distinct colon cancer tissues than other statistical approaches used heretofore. In addition, competing classification trees are displayed, which suggest that different genes may coregulate colon cancers.

Identification and classification of p53-regulated genes

Sequence-specific transactivation by p53 is essential to its role as a tumor suppressor. A modified tetracycline-inducible system was established to search for transcripts that were activated soon after p53 induction. Among 9,954 unique transcripts identified by serial analysis of gene expression, 34 were increased more than 10-fold; 31 of these had not previously been known to be regulated by p53. The transcription patterns of these genes, as well as previously described p53-regulated genes, were evaluated and classified in a panel of widely studied colorectal cancer cell lines. “Class I” genes were uniformly induced by p53 in all cell lines; “class II” genes were induced in a subset of the lines; and “class III” genes were not induced in any of the lines. These genes were also distinguished by the timing of their induction, their induction by clinically relevant chemotherapeutic agents, the absolute requirement for p53 in this induction, and their inducibility by p73, a p53 homolog. The results revealed substantial heterogeneity in the transcriptional responses to p53, even in cells derived from a single epithelial cell type, and pave the way to a deeper understanding of p53 tumor suppressor action.

Cytologically normal cells from neoplastic cervical samples display extensive structural abnormalities on IR spectroscopy: Implications for tumor biology

Fourier-transform IR (FT-IR) spectra of pelleted exfoliated cervical cells from patients with cervical cancer or dysplasia differ from those from normal women. To study the origin of these spectral changes, we obtained the FT-IR spectra of individual cervical cells from normal, dysplastic, and malignant cervical samples. Ninety five percent of normal superficial and intermediate cells displayed two distinct spectral patterns designated A and B, and 5% displayed an intermediate pattern, suggesting extensive structural heterogeneity among these cells. Parabasal and endocervical cells showed pattern B spectra. The spectra of malignant, dysplastic, and other abnormal cells also were characterized. Analysis of FT-IR spectra of over 2,000 individual cells from 10 normal females, 7 females with dysplasia, and 5 females with squamous cell carcinoma revealed that the spectra of normal-appearing intermediate and superficial cells of the cervix from women with either dysplasia or cancer differed from those of normal women. Chemometric and classical spectroscopic analysis showed a continuum of changes paralleling the transition from normalcy to malignancy. These findings suggest that (i) the structural changes underlying the spectroscopic changes are involved in or are a product of cervical carcinogenesis and (ii) the neoplastic process may be more extensive than currently recognized with morphological criteria. This approach may be useful for the structural study of neoplasia and also may be of help in the diagnosis or classification of cervical disorders.