EDUCATION

DM cases. Hybridizations were performed using commercially available RREB1/ MYB/CCND1 probes and established diagnostic cutoffs. For literature review of test performance in conventional melanoma we performed pubmed searches in combination with manual review of references and tabulated the number of abnormal vs. tested cases for each probe and overall sensitivity comparisons. Routine test performance measures were calculated and statistical signifi cance was defi ned as P<0.05. Results: We performed a total of 123 hybridizations in 15 SM, 4 mixed and 14 DM cases. The assay was overall 88% sensitive (n=29 true positives). Although the sensitivity in DM was substantially lower (10/14=71% DM whereas 18/19=95% SM or 304/360=84% conventional melanoma), the differences did not reach statistical signifi cance (P-range=0.14-1.0; Chi-square). Sensitivity by individual probesets was RREB1 (24/32=75%), MYB (10/27=37%) and CCND1 (6/29=21%). Due to the relatively high sensitivity of RREB1, our results indicate that a consecutive FISH-testing algorithm (Figure 1) can drastically reduce the number of hybridizations (i.e., from n=123 to n=57). Conclusions: The triple FISH assay employing RREB1, MYB and CCND1 probe sets is highly sensitive (88%) in SM/DM. We provide evidence for a practically effi cient consecutive testing algorithm (Figure 1). Notably, the relatively high false negative rate in DM underscores the need for an additional reliable confi rmatory melanoma assay and emphasizes the biological differences in this melanoma subtype.