The use of TMAs as a strategy to validate and characterize candidate biomarkers is gaining widespread acceptance as an alternative to more traditional immunohistochemical methods. Less widely appreciated, is that TMAs also facilitate a rapid assessment of biomarker expression in larger cohorts of tissue and are thereby amenable to the development of statistical models based on structured immunohistochemistry information . In this regard it is noteworthy that while there is general acknowledgement of the prognostic value of the Gleason score there is also agreement on the limitations of morphology in the prediction of clinical outcome for individual patients. In part this limitation may be attributable to the considerable interobserver variability in the assignment of Gleason grade .
In this study, a 255 core TMA comprising the spectrum of low grade to high grade Gleason scores of organ confined tumors was used to address several questions of clinical and pathological interest based on statistical modeling of the expression of selected biomarkers. These biomarkers were comprised of well characterized cell cycle (p53, p21, mdm2) and cell death (p53, bax, bcl-xL, bcl-2, bin1, CD95, NFkB) regulators variably expressed by human prostate cancers [5, 10–16]. Available evidence suggests that at least several of these proteins may be of predictive value [15, 17–21]. While the scope of this study is focused on a small number of selected biomarkers, the analysis is based on nearly 7,000 data elements and highlights the critical issues of data management and distribution. A tissue array database (TAD) was, therefore, developed to facilitate the structuring, storage, and distribution of biomarker information. TAD consists of an Active Server Page web interfaced with a relational database that automates recording biomarker scores and links the structured data with clinical and pathologic information. TAD is an open source application that can be installed locally.
It is widely appreciated that Gleason score 7 prostate cancers in aggregate have an intermediate prognosis but are clinically heterogeneous despite their morphologic similarities and, conversely, a significant number of Gleason score 6 prostate cancers, despite their overall good prognosis, exhibit clinically aggressive behavior . Clearly, a biomarker signature capable of contributing to the predictive value of accepted histopathologic information would be of considerable value. As a first step to realizing this goal we modeled expression of selected proteins on a tissue microarray comprising the spectrum of Gleason score organ confined tumors. Our analysis suggests that four markers (bcl-xL, Fas, bax, and p21) provided potentially important information predictive of clinically aggressive behavior. The findings also indicate, that tumors of equivalent Gleason grade although morphologically similar, have varying patterns of biomarker expression that correlate with the Gleason score of the tumor.
Represented in the TMA were 15 cases of Gleason score 6 (3 + 3) carcinoma consisting of tumors of transition zone (5 cases) or peripheral zone (10 cases) origin defined by established histopathological criteria, that enabled an assessment of the potential variability of individual biomarker expression as a function of zonal origin of the tumor. Tumors of peripheral zone origin that exhibited secondary involvement were excluded. It is noteworthy that bax, bin1, mdm2, p21, and p65 NFkB were all expressed at significantly higher levels in peripheral zone tumors compared to transition zone tumors of equivalent Gleason score. Although establishing the biological, or potential clinical, significance of these observations is beyond the scope of the current study it is of interest that transition zone prostate carcinomas are both less common and typically indolent compared to their counterparts arising from the peripheral zone [23, 24]. Additionally, it has been speculated that transition zone and peripheral zone carcinoma may arise from different precursor lesions . Although it has not been previously established, it is, therefore, not completely unanticipated that carcinomas arising from the peripheral zone would exhibit patterns of biomarker expression that differ from those arising in the transition zone.
In pairwise comparisons of the expression of the nine biomarkers using Spearman correlation coefficients several combinations of biomarkers exhibited highly concordant expression. There were significant associations between bcl-xL and bax, bax and bin1, and bax and p21. Although proapoptotic and antiapoptotic proteins of the bcl-2 gene family have been shown to be coordinately regulated , as well as bax and p21 , an association between bax and bin1 has not been described. It would be of interest to determine the extent to which apoptosis, or tumor suppressor activity, mediated by bin1 may be dependent on bax.
The results of this study suggest opportunities for future laboratory-based mechanistic studies. Finally, biostatistical strategies are presented that suggest it may be possible to use a panel of biomarkers to provide predictive information about Gleason score 7 prostate cancers. Ultimately, these findings will need to be expanded to determine, prospectively, whether they will prove of clinical utility beyond current histopathologic variables including Gleason grade and tumor stage. It is anticipated that multiparametric analyses can be performed on limited tumor specimens and will lead to improved clinical investigation and better validation of mechanistic concepts. Incremental advances in prostate cancer therapy will likely rely on integration of validated molecular information with the widely applied morphologic characterization of prostate cancer. Integration of this information with other prognostic variables including tumor stage and pretreatment PSA levels will be necessary to establish its ultimate utility. This will necessitate rigorous validation in clinically annotated specimens with long-term follow up.