We appreciate the work that Micci et al undertook with the aim of independently evaluating our finding that the ESRRA-C11orf20 is a recurrent gene fusion in serous ovarian carcinoma. In our original report, we found that 10/67 cases of serous ovarian carcinoma were positive in a nested PCR screen for the fusion transcript. As described in our 2011 manuscript, the estimated incidence rate was derived from observing fusion positive tumors at a rate of 1/25 among samples from the British Columbia Cancer Agency (BCCA) Tumour Tissue Repository screened in Victoria, BC and a rate of 9/42 samples from the Pacific Ovarian Cancer Research Consortium (POCRC) screened at the Fred Hutchinson Cancer Research Center (FHCRC).

Since the publication of our 2011 paper, we have not undertaken further screening of serous ovarian cancer samples using PCR. However, we have investigated RNA-Seq datasets from serous ovarian cancer generated by the TCGA consortium to test for the prevalence of the ESRRA-C11orf20 fusion. By surveying 420 samples, we detected evidence of fusion reads across the ESRRA-C11orf20 exon-exon junction in two tumors, and a third tumor with evidence of a fusion between ESRRA and KCNK4 at annotated exon boundaries (KCNK4 and C11orf20 share the same annotated 3' exons). These findings would produce a point estimate of the prevalence of the fusion transcript at 3/420 tumors, outside of the 95% confidence interval we reported in 2011. However, this estimate of incidence, 95% exact binomial confidence interval (0.001,0.02) overlaps the a 95% exact binomial confidence interval (0.001,0.20) for the incidence of the fusion derived from tumors screened in Victoria, BC.

The negative results of Micci et al’s PCR screen for the ESRRA-C11orf20 fusion raise legitimate questions about its prevalence and pathogenic significance. In our original report, we used nested PCR to detect fusion transcripts because we had evidence that our PCR assays were at the limit of detection of the fusion transcript in many tumors. This means that starting RNA input to an RT, the efficiency of the RT and concentration of the templates will affect subsequent attempts to detect the fusion transcript by RT-PCR, and that detection by RT-PCR and by RNA-Seq are likely to produce different estimates of the fusion prevalence. The lack of evidence for the fusion based on Micci et al’s RNA sequencing analysis of 23 ovarian cancers, only 10 of which were serous, is of uncertain relevance, as it is entirely consistent with the prevalence we reported.

Although we and our collaborators took great pains to avoid any possibility of contamination producing false-positive PCR results, including analyzing independent sets of tumors at geographically separated sites without any transfer of primers or DNA samples between the sites, we recognize the possibility that our estimate of the prevalence of the ESRRA-C11orf20 fusion may have been inflated by false positives due to cross-contamination. Regardless of the reason for discrepant estimates of prevalence, our findings, together with the data presented by Micci et al, suggest that the while the ESRRA-C11orf20 fusion is recurrent, the population wide incidence of detectable levels of the ESRRA-C11orf20 fusion in serous ovarian cancer is almost certainly lower than we estimated in our 2011 report.