The ability to parse tumors into subsets based on biomarker expression has many clinical applications; however, there is no global way to visualize the best cut-points for creating such divisions. We have developed a graphical method, the X-tile plot that illustrates the presence of substantial tumor subpopulations and shows the robustness of the relationship between a biomarker and outcome by construction of a two dimensional projection of every possible subpopulation. We validate X-tile plots by examining the expression of several established prognostic markers (human epidermal growth factor receptor-2, estrogen receptor, p53 expression, patient age, tumor size, and node number) in cohorts of breast cancer patients and show how X-tile plots of each marker predict population subsets rooted in the known biology of their expression.

The recent development of tissue microarrays-composed of hundreds of tissue sections from different tumors arrayed on a single glass slide-facilitates rapid evaluation of large-scale outcome studies. Realization of this potential depends on the ability to rapidly and precisely quantify the protein expression within each tissue spot. We have developed a set of algorithms that allow the rapid, automated, continuous and quantitative analysis of tissue microarrays, including the separation of tumor from stromal elements and the sub-cellular localization of signals. Validation studies using estrogen receptor in breast carcinoma show that automated analysis matches or exceeds the results of conventional pathologist-based scoring. Automated analysis and sub-cellular localization of beta-catenin in colon cancer identifies two novel, prognostically significant tumor subsets, not detected by traditional pathologist-based scoring. Development of automated analysis technology empowers tissue microarrays for use in discovery-type experiments (more typical of cDNA microarrays), with the added advantage of inclusion of long-term demographic and patient outcome information.

Using this system for classification, we define the molecular profile of HPV+ OSCC with favorable prognosis, namely HPV+/p16 high (class III). This study defines a novel classification scheme that may have value for patient stratification for clinical trials testing HPV-targeted therapies.

SUMMARY:The recent development of tissue microarray technology has potentiated large-scale retrospective cohort studies using archival formalin-fixed, paraffin-embedded tissues. A major obstacle to broad acceptance of microarrays is that they reduce the amount of tissue analyzed from a whole tissue section to a disk, 0.6 mm in diameter, that may not be representative of the protein expression patterns of the entire tumor. In this study, we examine the number to disks required to adequately represent the expression of three common antigens in invasive breast carcinoma-estrogen receptor, progesterone receptor, and the Her2/neu oncogene-in 38 cases of invasive breast carcinoma. We compared the staining of 2 to 10 microarray disks and the whole tissue sections from which they were derived and determined that analysis of two disks is comparable to analysis of a whole tissue section in more than 95% of cases. To evaluate the potential for using archival tissue in such arrays, we created a breast cancer microarray of 8 to 11 cases from each decade beginning in 1932 to the present day and evaluated the antigenicity of these markers and others. This array demonstrates that many proteins retain their antigenicity for more than 60 years, thus validating their study on archival tissues. We conclude that the tissue microarray technique, with 2-fold redundancy, is a valuable and accurate method for analysis of protein expression in large archival cohorts. (Lab Invest 2000, 80:1943-1949.

The heat shock protein HSP90 chaperones proteins implicated in breast cancer progression, including Her2/neu. HSP90-targeting agents are in clinical trials for breast cancer. HSP90 expression is high in breast cancer cell lines, yet no large studies have been conducted on expression in human tumors and the association with clinical/pathologic variables. Tissue microarrays containing 10 cell lines and primary specimens from 655 patients with 10-year follow-up were assessed using our automated quantitative analysis (AQUA) method; we used cytokeratin to define pixels as breast cancer (tumor mask) within the array spot and measured HSP90 expression within the mask using Cy5-conjugated antibodies. We similarly assessed estrogen receptor, progesterone receptor, and Her2/ neu expression. HSP90 expression was more variable in human tumors than in cell lines (P < 0.0001). High HSP90 expression was associated with decreased survival (P = 0.0024). On multivariable analysis, high HSP90 expression remained an independent prognostic marker. High HSP90 expression was associated with high Her2/neu and estrogen receptor, large tumors, high nuclear grade, and lymph node involvement. Although HSP90 levels were high in all our cell lines, expression in tumors was more variable. High HSP90 expression in primary breast cancer defines a population of patients with decreased survival. Evaluation of HSP90 expression in early-stage breast cancer may identify a subset of patients requiring more aggressive or pathway-targeted treatment. Prospective studies are needed to confirm the prognostic role of HSP90, as well as the predictive role of HSP90 expression in patients treated with HSP90 inhibitors.