Gene expression signatures as a guide to treatment strategies for in-transit metastatic melanoma.

9077 Background: In an era of targeted therapeutics biopsies are increasingly used to personalize cancer treatment. The utility of a single biopsy from a melanoma patient with multifocal disease to define the characteristics of that patient's tumor is unclear. The aim of this study was to evaluate the genetic and molecular relationship of multifocal lesions and determine if one lesion is representative of residual tumor burden. METHODS: Expression across 38k genes was measured using GeneChip microarrays. Binary regression, unsupervised hierarchical clustering, principal component analysis (PCA) and analysis of variance (ANOVA) were used to evaluate the patterns of expression across lesions. Predictions of sensitivity to chemotherapy and oncogenic signaling pathway activation were evaluated using signatures derived from the NCI-60 panel of cancer cell lines. RESULTS: 43 in-transit lesions were obtained from 17 different patients with multifocal extremity melanoma. PCA and unsupervised hierarchical clustering showed distinctly different patterns of gene expression across patients but significantly correlated (average r = 0.979) patterns within a patient. Sensitivity to melphalan and temozolomide was evaluated using defined gene signatures. Predictions across multifocal lesions were significantly correlated within an individual patient (ANOVA p-value: <0.001) and significantly different across patients (p<0.05). Similarly, predicted patterns of gene expression across 6 signaling pathways (Src, PI3K, Ras, Myc, E2F and β-catenin) were significantly correlated within but not across patients (p<0.0001 to 0.0006). CONCLUSIONS: Individual melanoma tumor nodules in patients with multifocal disease are similar and can be used to predict chemosensitivity, genetically characterize a patient's disease and evaluate the activation status of oncogenic signaling pathways. These findings will facilitate the application of microarray-based gene expression profiling in clinical melanoma trials, allowing for a more rational way to identify candidates for specific targeted therapies, and demonstrate that single biopsies obtained for correlative science studies are representative of residual tumor burden. No significant financial relationships to disclose.

Duke Scholars

Author Sin-Ho Jung Biostatistics & Bioinformatics, Division of Integrative Geno ...

Author Jin Soo Yoo Surgery, Minimally Invasive Surgery

Author Francis Ali-Osman Neurosurgery, Neuro-Oncology