Melanoma Signatures Come of Age

Home>
Specialty Care>
Dermatology>
Summary and Comment

Melanoma Signatures Come of Age

Gene expression profiling identified predictors of outcome in melanoma.

Global gene expression profiling of primary human melanomas is hampered by a lack of frozen tumors for analysis. Soon after its inception in the early 1980s, the European Organization for Research and Treatment of Cancer (EORTC) made a concerted effort to freeze representative portions of primary and metastatic melanomas. Twenty years later, this effort has paid off, as the large cancer consortium recently published the most definitive gene expression analysis of primary melanomas to date.

The study population consisted of 83 patients who were previously treated for primary melanoma and had available fresh-frozen tissue samples and follow-up information (median follow-up, 59 months). Using the Agilent whole human genome array, analysis based on similarities in expression identified two major subsets of these melanomas — 49 tumors with a median thickness of 1.7 mm formed the "left" cluster, and 33 tumors with a median thickness of 5.5 mm formed the "right" cluster (one tumor did not fit either). The between-cluster difference in median lesion thickness was statistically significant (P<0.001). When tumor subsets were matched with clinical outcomes, the left cluster was associated with significantly better 4-year survival rates than the right cluster (overall 4-year survival rate, 0.69 vs. 0.53, P=0.041).

The investigators used a 254-gene signature to assign tumors to the favorable or the unfavorable cluster. The protein expression of 23 of the 254 genes was analyzed immunohistochemically. Increased expression of eight of these proteins was significantly associated with favorable distant-metastasis–free and overall survival. In an independent set of 176 primary melanomas, immunoreactivity in two genes — MCM4 and MCM6 — was found to be significantly associated with survival after adjustment for tumor thickness, ulceration, age, and sex.

Comment: This consortial effort is a considerable achievement in personalized melanoma medicine. Identification and validation of a small group of expression signatures that correlate with survival is, in some way, the goal of many seekers of novel prognostic biomarkers. Taking a coherent, methodical approach to cancer prediction, the EORTC obtained what is probably the most robust data to date, and they have validated initial findings in independent sets. However, what does it all mean?

The tumor subsets segregated initially by thickness, which would explain prognostic differences. The authors also identified two genes from tens of thousands that predict outcome independent of thickness — a herculean effort, but are these markers clinically necessary when thickness and ulceration suffice in most patients? More than a marker that correlates with thickness, the field needs a marker of aggression in thin tumors. Another major challenge for microarray studies is validation by other groups: Subtle technologic differences between microarrays may not yield similar gene profiles. Despite these limitations, this impressive application of genomic technology will soon generate many testable hypotheses and translatable initiatives.

— Hensin Tsao, MD, PhD

Published in Journal Watch Dermatology July 28, 2006