To the editor: We read with interest the article by Lapointe et al,1 who reported a novel fusion of variant TMPRSS2 isoform with ERG, resulting in a variant TMPRSS2 (isoforms 2)-ERG fusion in prostate cancer. This finding has important implications for accurate detection of TMPRSS2-ERG fusion transcript, since 10% of specimens in their series expressed only the novel variant isoforms TMPRSS2-ERG fusion, which could be missed using the published RT-PCR assays for the known TMPRSS2 transcript.2, 3, 4, 5, 6 This is a particularly important issue for the development of PCR-based diagnostic tests. For tissue-based FISH assays, the two TMPRSS2 isoforms cannot be distinguished using the standard ERG break-apart assay.2, 4, 6

There appears to be some possibility of confusion regarding the ERG break-apart assays presented in their paper. The FISH images presented in Figure 3a are described as negative for the TMPRSS2-ERG fusion. Because TMPRSS2 and ERG are located so close together on 21q, all non-rearranged alleles should be immediately adjacent to each other (or even overlapping). In this figure, it appears more consistent with a break-apart signal pattern, suggesting gene fusion. This may be due to overlapping nuclei, making it unclear which signals belong to which nucleus, and whether all signals have been captured,as they may not all be on the same plane. Figure 3b demonstrates very well the TMPRRS2-ERG rearrangement through deletion. There is no significant visible difference between the relative signal positions of the non-rearranged allele and the allele with the rearrangement. As the signal distance is not informative, the rearrangement only becomes obvious because of the missing green signal between the red and blue signals. The signal patterns in Figure 3c are equally confusing. As in Figure 3b, the rearrangement is most clearly shown by the absence of the green signals (indicating fusion through deletion). However, the signals for the non-rearranged allele in the upper-right nucleus seem to be too distant to be from the same nucleus. In Figure 3b and the other nuclei in Figure 3c, the non-rearranged aqua and red signals are immediately adjacent to each other. Also, there appears to be an additional non-rearranged allele in the large nucleus in the bottom center. We believe that these inconsistencies are due to overlapping nuclei as shown in Figure 3a.

We believe it is important to point out the ambiguity of the images to avoid possible confusion for the reader and in interpreting future TMPRSS2-ERG FISH assays.