Gene expression analysis has also been used to identify

Gene expression analysis has also been used to identify signatures that may predict which patients are most likely to need or benefit from systemic chemotherapy. Smith et al. [63] reported a 20-gene signature that predicts for risk of occult nodal involvement at the time of cystectomy. In a validation cohort, the relative risk of nodal involvement in the defined high- and low-risk groups was 1.74 and 0.70, respectively. These findings are intriguing, though illustrate a common challenge of how best to define clinically meaningful thresholds often posed by the use of gene expression analysis for biomarker discovery. For example, what relative risk of nodal involvement should be acceptable to influence clinical recommendations? The answer to this and similar questions ultimately requires incorporation of gene expression data into prospective clinical trials before routine use could be expected.
Fortunately, efforts to develop clinical trials based on molecular profiling are underway. The Co-eXpression ExtrapolatioN (COXEN) algorithm is an interesting application of gene expression analysis to predict chemosensitivity in patients with chemokine receptor cancer [64]. This algorithm uses gene expression profiles generated from the U.S. National Cancer Institute\’s effort to screen more than 100,000 chemical compounds against a panel of 60 human tumor cell lines and matches profiles with clinical specimens to predict chemosensitivity. When this algorithm was applied to published bladder cancer gene expression sets, the COXEN score effectively predicted for chemotherapy response using multivariate analysis [65]. The algorithm serves as the basis for a planned cooperative group study to determine whether COXEN analysis of patient specimens can identify those most likely to respond to neoadjuvant cisplatin-based chemotherapy.
As the use of molecular profiling in bladder cancer increases, development of methodologies that can use routine clinical specimens will be critical. Most molecular profiling study results in bladder cancer have been derived from fresh frozen tissue; however, the use of formalin-fixed, paraffin embedded (FFPE) specimens is more practical for routine clinical use. As an example, the use of molecular inversion probe assays for identification of genomic changes and mutational profiling from FFPE specimens has recently been reported. Using this approach, Chekaluk and colleagues [66] identified 44 regions of chromosomal copy number changes and further examined 9 regions of amplification, based on the presumption that amplified genomic regions may be enriched for driver oncogenes. Using multiplex ligation-dependent probe assays for validation, the investigators identified the following genomic regions (and genes) as being most frequently amplified in primary bladder tumors: 1q23.3 (multiple genes), 6p22.3 (E2F3-SOX4), and 11q13.3 (CCND1). Rare amplification of 17q12 (ErbB2) was also noted. This work is notable for providing additional evidence of putative oncogenes that may be amenable to therapeutic targeting (CDK4 inhibitors for CCND1-amplified tumors and anti-her2 agents for ErbB2-amplified tumors) and also for describing a novel approach for performing genomic analysis on specimens derived from FFPE tissue.
Molecular concordance between primary tumors and metastatic lesions is not well defined in bladder cancer, though emerging data suggest discordance may be prevalent. Using molecular inversion probe assays on 30 primary tumors and 33 metastatic lesions (with 14 paired samples from the same patients), increased genomic copy number changes were noted in metastatic lesions compared with primary tumors (8.4% vs. 4.3%) [67]. Primary and metastatic site copy number gains were present in E2F3 (7% vs. 27%) and in CCND1 (7% vs. 18%) genes, respectively. ErbB2 amplifications were present in 7% of primary lesions and 12% of metastatic lesions, and the investigators noted in the paired specimens that 7 of 14 patients contained genomic aberrations in the metastatic lesion not found in the primary tumor. If confirmed with larger data sets, these findings have a striking potential to affect clinical management. In the future, patients with metastatic disease may conceivably need biopsies of multiple lesions to assess for molecular heterogeneity that could influence the optimal selection of targeted agents.