Towards improved biomarker studies of cervical neoplasia: effects of precolposcopic procedures on gene expression patterns.

Among tumor sites, cervical cancer offers an ideal model for investigating differences in gene expression associated with transitions from normal to precancer and invasion to cancer. To evaluate the validity of assessing gene expression in cervical tissues acquired in a clinical setting, we investigated whether standard procedures, namely the application of acetic acid and/or Lugol's iodine, employed for the visualization of colposcopically directed biopsies, altered patterns in oligonucleotide (oligo) arrays. We compared microarray profiles from six women, each with three adjacent tissue samples removed from benign hysterectomy specimens and treated as follows: immediately frozen, acetic acid application only, acetic acid, and Lugol's iodine. Of the 22,464 original spots on the microarray, 4,850 spots were expressed at detectable levels for further evaluation upon data normalization and filtration. For each spot, the difference between topical applications was computed, and P values were calculated using a bivariate T2 test. Upon adjustment for multiple comparisons using both the Holm's and Hochberg's procedures as well as the False Discovery Rate (Benjamini-Hochberg and Benjamini-Yeuketili [BY]), we failed to identify genes differentially expressed and conclude that standard precolposcopic procedures do not substantially affect the overall gene expression patterns in the normal cervix.